diff --git a/.DS_Store b/.DS_Store
new file mode 100644
index 0000000..8aee692
Binary files /dev/null and b/.DS_Store differ
diff --git a/ai-templates-0/README.md b/ai-templates-0/README.md
new file mode 100644
index 0000000..b0db4dd
--- /dev/null
+++ b/ai-templates-0/README.md
@@ -0,0 +1,187 @@
+# Portainer AI Templates (v2)
+
+> **26 production-ready AI/ML Docker Compose stacks for Portainer** — filling the AI gap in the official v3 template library. Aligned with an AI infrastructure positioning strategy for Portainer.
+
+## Background
+
+The official [Portainer v3 templates](https://raw.githubusercontent.com/portainer/templates/v3/templates.json) contain **71 templates** with **zero pure AI/ML deployments**. This repository provides a curated, Portainer-compatible template set covering the entire AI infrastructure stack — from edge inference to distributed training to governed ML pipelines.
+
+See [docs/AI_GAP_ANALYSIS.md](docs/AI_GAP_ANALYSIS.md) for the full gap analysis.
+
+## Homepage Alignment
+
+These templates map directly to the AI infrastructure positioning pillars:
+
+| Mock-Up Pillar | Templates Covering It |
+|---|---|
+| **GPU-Aware Fleet Management** | Triton, vLLM, NVIDIA NIM, Ray Cluster, Ollama, LocalAI |
+| **Model Lifecycle Governance** | MLflow + MinIO (Production MLOps), Prefect, BentoML, Label Studio |
+| **Edge AI Deployment** | ONNX Runtime (CPU/edge profile), Triton, DeepStream |
+| **Self-Service AI Stacks** | Open WebUI, Langflow, Flowise, n8n AI, Jupyter GPU |
+| **LLM Fine-Tune** (diagram) | Ray Cluster (distributed training) |
+| **RAG Pipeline** (diagram) | Qdrant, ChromaDB, Weaviate + Langflow/Flowise |
+| **Vision Model** (diagram) | DeepStream, ComfyUI, Stable Diffusion WebUI |
+| **Anomaly Detection** (diagram) | DeepStream (video analytics), Triton (custom models) |
+
+## Quick Start
+
+### Option A: Use as Custom Template URL in Portainer
+
+1. In Portainer, go to **Settings > App Templates**
+2. Set the URL to:
+   ```
+   https://git.oe74.net/adelorenzo/portainer_scripts/raw/branch/master/ai-templates/portainer-ai-templates.json
+   ```
+3. Click **Save** — all 26 AI templates appear in your App Templates list
+
+### Option B: Deploy Individual Stacks
+
+```bash
+cd stacks/ollama
+docker compose up -d
+```
+
+## Template Catalog
+
+### LLM Inference and Model Serving
+
+| # | Template | Port | GPU | Description |
+|---|---|---|---|---|
+| 1 | **Ollama** | 11434 | Yes | Local LLM engine — Llama, Mistral, Qwen, Gemma, Phi |
+| 2 | **Open WebUI + Ollama** | 3000 | Yes | ChatGPT-like UI bundled with Ollama backend |
+| 3 | **LocalAI** | 8080 | Yes | Drop-in OpenAI API replacement |
+| 4 | **vLLM** | 8000 | Yes | High-throughput serving with PagedAttention |
+| 5 | **Text Gen WebUI** | 7860 | Yes | Comprehensive LLM interface (oobabooga) |
+| 6 | **LiteLLM Proxy** | 4000 | No | Unified API gateway for 100+ LLM providers |
+| 26 | **NVIDIA NIM** | 8000 | Yes | Enterprise TensorRT-LLM optimized inference |
+
+### Production Inference Serving
+
+| # | Template | Port | GPU | Description |
+|---|---|---|---|---|
+| 19 | **NVIDIA Triton** | 8000 | Yes | Multi-framework inference server (TensorRT, ONNX, PyTorch, TF) |
+| 20 | **ONNX Runtime** | 8001 | Optional | Lightweight inference with GPU and CPU/edge profiles |
+| 24 | **BentoML** | 3000 | Yes | Model packaging and serving with metrics |
+
+### Image and Video Generation
+
+| # | Template | Port | GPU | Description |
+|---|---|---|---|---|
+| 7 | **ComfyUI** | 8188 | Yes | Node-based Stable Diffusion workflow engine |
+| 8 | **Stable Diffusion WebUI** | 7860 | Yes | AUTOMATIC1111 interface for image generation |
+
+### Industrial AI and Computer Vision
+
+| # | Template | Port | GPU | Description |
+|---|---|---|---|---|
+| 21 | **NVIDIA DeepStream** | 8554 | Yes | Video analytics for inspection, anomaly detection, smart factory |
+
+### Distributed Training
+
+| # | Template | Port | GPU | Description |
+|---|---|---|---|---|
+| 22 | **Ray Cluster** | 8265 | Yes | Head + workers for LLM fine-tuning, distributed training, Ray Serve |
+
+### AI Agents and Workflows
+
+| # | Template | Port | GPU | Description |
+|---|---|---|---|---|
+| 9 | **Langflow** | 7860 | No | Visual multi-agent and RAG pipeline builder |
+| 10 | **Flowise** | 3000 | No | Drag-and-drop LLM chatflow builder |
+| 11 | **n8n (AI-Enabled)** | 5678 | No | Workflow automation with AI agent nodes |
+
+### Vector Databases
+
+| # | Template | Port | GPU | Description |
+|---|---|---|---|---|
+| 12 | **Qdrant** | 6333 | No | High-performance vector similarity search |
+| 13 | **ChromaDB** | 8000 | No | AI-native embedding database |
+| 14 | **Weaviate** | 8080 | No | Vector DB with built-in vectorization modules |
+
+### ML Operations and Governance
+
+| # | Template | Port | GPU | Description |
+|---|---|---|---|---|
+| 15 | **MLflow** | 5000 | No | Experiment tracking and model registry (SQLite) |
+| 25 | **MLflow + MinIO** | 5000 | No | Production MLOps: PostgreSQL + S3 artifact store |
+| 23 | **Prefect** | 4200 | No | Governed ML pipeline orchestration with audit logging |
+| 16 | **Label Studio** | 8080 | No | Multi-type data labeling platform |
+| 17 | **Jupyter (GPU/PyTorch)** | 8888 | Yes | GPU-accelerated notebooks |
+
+### Speech and Audio
+
+| # | Template | Port | GPU | Description |
+|---|---|---|---|---|
+| 18 | **Whisper ASR** | 9000 | Yes | Speech-to-text API server |
+
+## GPU Requirements
+
+Templates marked **GPU: Yes** require:
+- NVIDIA GPU with CUDA support
+- [NVIDIA Container Toolkit](https://docs.nvidia.com/datacenter/cloud-native/container-toolkit/install-guide.html) installed
+- Docker configured with `nvidia` runtime
+
+**Edge deployments (ONNX Runtime CPU profile):** No GPU required — runs on ARM or x86 with constrained CPU/memory limits.
+
+For AMD GPUs (ROCm), modify the `deploy.resources` section to use ROCm-compatible images and remove the NVIDIA device reservation.
+
+## File Structure
+
+```
+ai-templates/
+├── portainer-ai-templates.json       # Portainer v3 template definition (26 templates)
+├── README.md
+├── docs/
+│   └── AI_GAP_ANALYSIS.md           # Analysis of official templates gap
+└── stacks/
+    ├── ollama/                        # LLM Inference
+    ├── open-webui/
+    ├── localai/
+    ├── vllm/
+    ├── text-generation-webui/
+    ├── litellm/
+    ├── nvidia-nim/                    # v2: Enterprise inference
+    ├── triton/                        # v2: Production inference serving
+    ├── onnx-runtime/                  # v2: Edge-friendly inference
+    ├── bentoml/                       # v2: Model packaging + serving
+    ├── deepstream/                    # v2: Industrial computer vision
+    ├── ray-cluster/                   # v2: Distributed training
+    ├── prefect/                       # v2: Governed ML pipelines
+    ├── minio-mlops/                   # v2: Production MLOps stack
+    ├── comfyui/                       # Image generation
+    ├── stable-diffusion-webui/
+    ├── langflow/                      # AI agents
+    ├── flowise/
+    ├── n8n-ai/
+    ├── qdrant/                        # Vector databases
+    ├── chromadb/
+    ├── weaviate/
+    ├── mlflow/                        # ML operations
+    ├── label-studio/
+    ├── jupyter-gpu/
+    └── whisper/                       # Speech
+```
+
+## Changelog
+
+### v2 (March 2026)
+- Added 8 templates to close alignment gap with AI infrastructure positioning:
+  - **NVIDIA Triton Inference Server** — production multi-framework inference
+  - **ONNX Runtime Server** — lightweight edge inference with CPU/GPU profiles
+  - **NVIDIA DeepStream** — industrial computer vision and video analytics
+  - **Ray Cluster (GPU)** — distributed training and fine-tuning
+  - **Prefect** — governed ML pipeline orchestration
+  - **BentoML** — model packaging and serving
+  - **MLflow + MinIO** — production MLOps with S3 artifact governance
+  - **NVIDIA NIM** — enterprise-optimized LLM inference
+
+### v1 (March 2026)
+- Initial 18 AI templates covering LLM inference, image generation, agents, vector DBs, MLOps, and speech
+
+## License
+
+These templates reference publicly available Docker images from their respective maintainers. Each tool has its own license — refer to the individual project documentation.
+
+---
+
+*Portainer AI Templates by Adolfo De Lorenzo — March 2026*
diff --git a/ai-templates-0/docs/AI_GAP_ANALYSIS.md b/ai-templates-0/docs/AI_GAP_ANALYSIS.md
new file mode 100644
index 0000000..56beee0
--- /dev/null
+++ b/ai-templates-0/docs/AI_GAP_ANALYSIS.md
@@ -0,0 +1,89 @@
+# Portainer v3 Templates — AI Gap Analysis
+
+## Overview
+
+The official Portainer v3 templates (`templates.json`) contain **71 templates** across the following categories:
+
+| Category | Count | Examples |
+|---|---|---|
+| Database | 10 | MySQL, PostgreSQL, Mongo, Redis, CrateDB, Elasticsearch, CockroachDB, TimescaleDB |
+| Edge/IIoT | 14 | Softing EdgeConnectors, OPC Router, TOSIBOX, EMQX MQTT, Mosquitto, Node-RED, Litmus Edge |
+| Web/CMS | 8 | Nginx, Caddy, WordPress, Drupal, Joomla, Ghost, Plone |
+| DevOps/CI | 5 | Jenkins, GitLab CE, Dokku, Registry |
+| Monitoring | 4 | Grafana, Datadog, Sematext, Swarm Monitoring |
+| Messaging | 1 | RabbitMQ |
+| Storage | 3 | Minio, Scality S3, File Browser |
+| Serverless | 2 | OpenFaaS, IronFunctions |
+| Other | 6 | Ubuntu, NodeJS, Portainer Agent, OpenAMT, FDO, LiveSwitch |
+
+## AI Template Count in Official Repo: **0**
+
+There are **zero purely AI/ML-focused templates** in the current v3 template list.
+
+### Closest to AI
+
+- **Litmus Edge** (#70, #71) — Described as "enables industrial AI at scale" but is an OT data platform, not an AI deployment.
+- **Elasticsearch** (#13) — Used in vector search / RAG pipelines but is a general-purpose search engine.
+
+---
+
+## v2 Coverage Map
+
+This repository now provides **26 AI templates** organized into 9 sub-categories, mapped against the 4 AI infrastructure positioning pillars:
+
+### Pillar 1: GPU-Aware Fleet Management
+| Template | What It Proves |
+|---|---|
+| NVIDIA Triton | Multi-framework model serving across GPU fleet with dynamic batching |
+| vLLM | High-throughput LLM inference with tensor parallelism across GPUs |
+| NVIDIA NIM | Enterprise-grade NVIDIA-optimized inference microservices |
+| Ray Cluster | Distributed GPU scheduling across head + worker nodes |
+| Ollama / LocalAI | Single-node GPU inference engines |
+
+### Pillar 2: Model Lifecycle Governance
+| Template | What It Proves |
+|---|---|
+| MLflow + MinIO (Prod) | Versioned model registry + S3 artifact store + PostgreSQL tracking |
+| Prefect | Governed pipeline orchestration with scheduling, retries, audit logs |
+| BentoML | Model packaging with versioning and metrics endpoints |
+| Label Studio | Data labeling with project-level access control |
+| MLflow (standalone) | Experiment tracking and model comparison |
+
+### Pillar 3: Edge AI Deployment
+| Template | What It Proves |
+|---|---|
+| ONNX Runtime (edge profile) | CPU-only inference with memory/CPU limits for constrained devices |
+| NVIDIA Triton | Supports Jetson via multiarch images, model polling for OTA updates |
+| NVIDIA DeepStream | Video analytics pipeline for factory-floor cameras |
+
+### Pillar 4: Self-Service AI Stacks
+| Template | What It Proves |
+|---|---|
+| Open WebUI + Ollama | One-click ChatGPT-like deployment, no CLI needed |
+| Langflow / Flowise | Visual drag-and-drop agent builders |
+| n8n (AI-Enabled) | Workflow automation with AI nodes, accessible to non-developers |
+| Jupyter GPU | Notebook environment for data science teams |
+
+### Architecture Diagram Workloads
+| Diagram Node | Template(s) |
+|---|---|
+| LLM Fine-Tune | Ray Cluster |
+| RAG Pipeline | Qdrant + ChromaDB + Weaviate + Langflow/Flowise |
+| Vision Model | DeepStream, ComfyUI, Stable Diffusion WebUI |
+| Anomaly Detection | DeepStream (video analytics), Triton (custom ONNX/TensorRT models) |
+
+---
+
+## Remaining Gaps (Future Work)
+
+| Gap | Why It Matters | Potential Addition |
+|---|---|---|
+| ARM/Jetson-native images | True edge AI on embedded devices | Triton Jetson images, ONNX Runtime ARM builds |
+| Air-gapped deployment | Industrial environments with no internet | Offline model bundling scripts |
+| Model A/B testing | Production model governance | Seldon Core or custom Envoy routing |
+| Federated learning | Privacy-preserving distributed training | NVIDIA FLARE or Flower |
+| LLM evaluation/guardrails | Safety and quality governance | Ragas, DeepEval, NVIDIA NeMo Guardrails |
+
+---
+
+*Generated: March 2026 — For use with Portainer Business Edition and Community Edition*
diff --git a/ai-templates-0/portainer-ai-templates.json b/ai-templates-0/portainer-ai-templates.json
new file mode 100644
index 0000000..587c965
--- /dev/null
+++ b/ai-templates-0/portainer-ai-templates.json
@@ -0,0 +1,959 @@
+{
+  "version": "3",
+  "templates": [
+    {
+      "id": 1,
+      "type": 3,
+      "title": "Ollama",
+      "description": "Local LLM inference engine supporting Llama, Mistral, Qwen, Gemma, Phi and 100+ models with GPU acceleration",
+      "note": "Requires NVIDIA GPU with Docker GPU runtime configured. Pull models after deployment with: <code>docker exec ollama ollama pull llama3.1</code>",
+      "categories": ["ai", "llm", "inference"],
+      "platform": "linux",
+      "logo": "https://ollama.com/public/ollama.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/ollama/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "OLLAMA_PORT",
+          "label": "Ollama API port",
+          "default": "11434"
+        },
+        {
+          "name": "OLLAMA_NUM_PARALLEL",
+          "label": "Max parallel requests",
+          "default": "4"
+        },
+        {
+          "name": "OLLAMA_MAX_LOADED_MODELS",
+          "label": "Max models loaded in VRAM",
+          "default": "2"
+        }
+      ]
+    },
+    {
+      "id": 2,
+      "type": 3,
+      "title": "Open WebUI + Ollama",
+      "description": "Full-featured ChatGPT-like web interface bundled with Ollama backend for local LLM inference",
+      "note": "Access the web UI at the configured port. First user to register becomes admin. Requires NVIDIA GPU.",
+      "categories": ["ai", "llm", "chat-ui"],
+      "platform": "linux",
+      "logo": "https://docs.openwebui.com/img/logo.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/open-webui/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "OPEN_WEBUI_PORT",
+          "label": "Web UI port",
+          "default": "3000"
+        },
+        {
+          "name": "OLLAMA_PORT",
+          "label": "Ollama API port",
+          "default": "11434"
+        },
+        {
+          "name": "WEBUI_SECRET_KEY",
+          "label": "Secret key for sessions",
+          "default": "changeme"
+        },
+        {
+          "name": "ENABLE_SIGNUP",
+          "label": "Allow user registration",
+          "default": "true"
+        }
+      ]
+    },
+    {
+      "id": 3,
+      "type": 3,
+      "title": "LocalAI",
+      "description": "Drop-in OpenAI API compatible replacement. Run LLMs, generate images, audio locally with GPU acceleration",
+      "note": "Exposes an OpenAI-compatible API at /v1/. Models can be loaded via the API or placed in the models volume.",
+      "categories": ["ai", "llm", "openai-api"],
+      "platform": "linux",
+      "logo": "https://localai.io/logo.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/localai/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "LOCALAI_PORT",
+          "label": "API port",
+          "default": "8080"
+        },
+        {
+          "name": "THREADS",
+          "label": "CPU threads for inference",
+          "default": "4"
+        },
+        {
+          "name": "CONTEXT_SIZE",
+          "label": "Default context window size",
+          "default": "4096"
+        }
+      ]
+    },
+    {
+      "id": 4,
+      "type": 3,
+      "title": "vLLM",
+      "description": "High-throughput LLM serving engine with PagedAttention, continuous batching, and OpenAI-compatible API",
+      "note": "Requires NVIDIA GPU with sufficient VRAM for the chosen model. HuggingFace token needed for gated models.",
+      "categories": ["ai", "llm", "inference", "high-performance"],
+      "platform": "linux",
+      "logo": "https://docs.vllm.ai/en/latest/_static/vllm-logo-text-light.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/vllm/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "VLLM_PORT",
+          "label": "API port",
+          "default": "8000"
+        },
+        {
+          "name": "MODEL_NAME",
+          "label": "HuggingFace model ID",
+          "default": "meta-llama/Llama-3.1-8B-Instruct"
+        },
+        {
+          "name": "HF_TOKEN",
+          "label": "HuggingFace access token"
+        },
+        {
+          "name": "MAX_MODEL_LEN",
+          "label": "Max sequence length",
+          "default": "4096"
+        },
+        {
+          "name": "GPU_MEM_UTIL",
+          "label": "GPU memory utilization (0-1)",
+          "default": "0.90"
+        },
+        {
+          "name": "TENSOR_PARALLEL",
+          "label": "Tensor parallel GPU count",
+          "default": "1"
+        }
+      ]
+    },
+    {
+      "id": 5,
+      "type": 3,
+      "title": "Text Generation WebUI",
+      "description": "Comprehensive web UI for running LLMs locally (oobabooga). Supports GGUF, GPTQ, AWQ, EXL2, and HF formats",
+      "note": "Requires NVIDIA GPU. Models should be placed in the models volume. Supports extensions for RAG, TTS, and more.",
+      "categories": ["ai", "llm", "chat-ui"],
+      "platform": "linux",
+      "logo": "https://raw.githubusercontent.com/oobabooga/text-generation-webui/main/docs/logo.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/text-generation-webui/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "WEBUI_PORT",
+          "label": "Web UI port",
+          "default": "7860"
+        },
+        {
+          "name": "API_PORT",
+          "label": "API port",
+          "default": "5000"
+        },
+        {
+          "name": "STREAM_PORT",
+          "label": "Streaming API port",
+          "default": "5005"
+        },
+        {
+          "name": "EXTRA_LAUNCH_ARGS",
+          "label": "Extra launch arguments",
+          "default": "--listen --api"
+        }
+      ]
+    },
+    {
+      "id": 6,
+      "type": 3,
+      "title": "LiteLLM Proxy",
+      "description": "Unified LLM API gateway supporting 100+ providers (OpenAI, Anthropic, Ollama, vLLM, etc.) with spend tracking and load balancing",
+      "note": "Configure models in /app/config/litellm_config.yaml after deployment. Includes PostgreSQL for usage tracking.",
+      "categories": ["ai", "llm", "api-gateway", "proxy"],
+      "platform": "linux",
+      "logo": "https://litellm.ai/favicon.ico",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/litellm/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "LITELLM_PORT",
+          "label": "Proxy API port",
+          "default": "4000"
+        },
+        {
+          "name": "LITELLM_MASTER_KEY",
+          "label": "Master API key",
+          "default": "sk-master-key"
+        },
+        {
+          "name": "PG_USER",
+          "label": "PostgreSQL user",
+          "default": "litellm"
+        },
+        {
+          "name": "PG_PASSWORD",
+          "label": "PostgreSQL password",
+          "default": "litellm"
+        }
+      ]
+    },
+    {
+      "id": 7,
+      "type": 3,
+      "title": "ComfyUI",
+      "description": "Node-based Stable Diffusion workflow engine for image and video generation with GPU acceleration",
+      "note": "Requires NVIDIA GPU. Access the node editor at the configured port. Models go in the models volume.",
+      "categories": ["ai", "image-generation", "stable-diffusion"],
+      "platform": "linux",
+      "logo": "https://raw.githubusercontent.com/comfyanonymous/ComfyUI/master/web/assets/comfyui-logo.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/comfyui/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "COMFYUI_PORT",
+          "label": "Web UI port",
+          "default": "8188"
+        },
+        {
+          "name": "CLI_ARGS",
+          "label": "Launch arguments",
+          "default": "--listen 0.0.0.0 --port 8188"
+        }
+      ]
+    },
+    {
+      "id": 8,
+      "type": 3,
+      "title": "Stable Diffusion WebUI",
+      "description": "AUTOMATIC1111 web interface for Stable Diffusion image generation with extensive extension ecosystem",
+      "note": "Requires NVIDIA GPU with 8GB+ VRAM. First startup downloads the base model and may take several minutes.",
+      "categories": ["ai", "image-generation", "stable-diffusion"],
+      "platform": "linux",
+      "logo": "https://raw.githubusercontent.com/AUTOMATIC1111/stable-diffusion-webui/master/html/logo.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/stable-diffusion-webui/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "SD_PORT",
+          "label": "Web UI port",
+          "default": "7860"
+        },
+        {
+          "name": "CLI_ARGS",
+          "label": "Launch arguments",
+          "default": "--listen --api --xformers"
+        }
+      ]
+    },
+    {
+      "id": 9,
+      "type": 3,
+      "title": "Langflow",
+      "description": "Visual framework for building multi-agent and RAG applications. Drag-and-drop LLM pipeline builder",
+      "note": "Access the visual editor at the configured port. Connect to Ollama, OpenAI, or any LLM backend.",
+      "categories": ["ai", "agents", "rag", "workflows"],
+      "platform": "linux",
+      "logo": "https://avatars.githubusercontent.com/u/128686189",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/langflow/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "LANGFLOW_PORT",
+          "label": "Web UI port",
+          "default": "7860"
+        },
+        {
+          "name": "AUTO_LOGIN",
+          "label": "Skip login screen",
+          "default": "true"
+        }
+      ]
+    },
+    {
+      "id": 10,
+      "type": 3,
+      "title": "Flowise",
+      "description": "Drag-and-drop LLM orchestration tool. Build chatbots, agents, and RAG pipelines without coding",
+      "note": "Default credentials are admin/changeme. Connect to any OpenAI-compatible API backend.",
+      "categories": ["ai", "agents", "rag", "chatbots"],
+      "platform": "linux",
+      "logo": "https://flowiseai.com/favicon.ico",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/flowise/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "FLOWISE_PORT",
+          "label": "Web UI port",
+          "default": "3000"
+        },
+        {
+          "name": "FLOWISE_USERNAME",
+          "label": "Admin username",
+          "default": "admin"
+        },
+        {
+          "name": "FLOWISE_PASSWORD",
+          "label": "Admin password",
+          "default": "changeme"
+        }
+      ]
+    },
+    {
+      "id": 11,
+      "type": 3,
+      "title": "n8n (AI-Enabled)",
+      "description": "Workflow automation platform with built-in AI agent nodes, LLM chains, and vector store integrations",
+      "note": "AI features include: AI Agent nodes, LLM Chain, Document Loaders, Vector Stores, Text Splitters, and Memory nodes.",
+      "categories": ["ai", "automation", "workflows", "agents"],
+      "platform": "linux",
+      "logo": "https://n8n.io/favicon.ico",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/n8n-ai/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "N8N_PORT",
+          "label": "Web UI port",
+          "default": "5678"
+        },
+        {
+          "name": "N8N_USER",
+          "label": "Admin username",
+          "default": "admin"
+        },
+        {
+          "name": "N8N_PASSWORD",
+          "label": "Admin password",
+          "default": "changeme"
+        },
+        {
+          "name": "WEBHOOK_URL",
+          "label": "External webhook URL",
+          "default": "http://localhost:5678/"
+        }
+      ]
+    },
+    {
+      "id": 12,
+      "type": 3,
+      "title": "Qdrant",
+      "description": "High-performance vector similarity search engine for RAG, semantic search, and AI applications",
+      "note": "REST API on port 6333, gRPC on 6334. Supports filtering, payload indexing, and distributed mode.",
+      "categories": ["ai", "vector-database", "rag", "embeddings"],
+      "platform": "linux",
+      "logo": "https://qdrant.tech/images/logo_with_text.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/qdrant/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "QDRANT_HTTP_PORT",
+          "label": "REST API port",
+          "default": "6333"
+        },
+        {
+          "name": "QDRANT_GRPC_PORT",
+          "label": "gRPC port",
+          "default": "6334"
+        },
+        {
+          "name": "QDRANT_API_KEY",
+          "label": "API key (optional)"
+        }
+      ]
+    },
+    {
+      "id": 13,
+      "type": 3,
+      "title": "ChromaDB",
+      "description": "AI-native open-source embedding database. The easiest vector store to get started with for RAG applications",
+      "note": "Persistent storage enabled by default. Compatible with LangChain, LlamaIndex, and all major AI frameworks.",
+      "categories": ["ai", "vector-database", "rag", "embeddings"],
+      "platform": "linux",
+      "logo": "https://www.trychroma.com/chroma-logo.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/chromadb/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "CHROMA_PORT",
+          "label": "API port",
+          "default": "8000"
+        },
+        {
+          "name": "CHROMA_TOKEN",
+          "label": "Auth token (optional)"
+        },
+        {
+          "name": "TELEMETRY",
+          "label": "Anonymous telemetry",
+          "default": "FALSE"
+        }
+      ]
+    },
+    {
+      "id": 14,
+      "type": 3,
+      "title": "Weaviate",
+      "description": "AI-native vector database with built-in vectorization modules and hybrid search capabilities",
+      "note": "Supports text2vec-transformers, generative-openai, and many other modules. Configure modules via environment variables.",
+      "categories": ["ai", "vector-database", "rag", "search"],
+      "platform": "linux",
+      "logo": "https://weaviate.io/img/site/weaviate-logo-light.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/weaviate/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "WEAVIATE_HTTP_PORT",
+          "label": "HTTP API port",
+          "default": "8080"
+        },
+        {
+          "name": "WEAVIATE_GRPC_PORT",
+          "label": "gRPC port",
+          "default": "50051"
+        },
+        {
+          "name": "VECTORIZER",
+          "label": "Default vectorizer module",
+          "default": "none"
+        },
+        {
+          "name": "MODULES",
+          "label": "Enabled modules",
+          "default": "text2vec-transformers,generative-openai"
+        },
+        {
+          "name": "ANON_ACCESS",
+          "label": "Anonymous access enabled",
+          "default": "true"
+        }
+      ]
+    },
+    {
+      "id": 15,
+      "type": 3,
+      "title": "MLflow",
+      "description": "Open-source ML lifecycle platform — experiment tracking, model registry, and model serving",
+      "note": "Access the tracking UI at the configured port. Uses SQLite backend by default — switch to PostgreSQL for production.",
+      "categories": ["ai", "mlops", "experiment-tracking", "model-registry"],
+      "platform": "linux",
+      "logo": "https://mlflow.org/img/mlflow-black.svg",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/mlflow/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "MLFLOW_PORT",
+          "label": "Tracking UI port",
+          "default": "5000"
+        }
+      ]
+    },
+    {
+      "id": 16,
+      "type": 3,
+      "title": "Label Studio",
+      "description": "Multi-type data labeling and annotation platform for training ML and AI models",
+      "note": "Supports image, text, audio, video, and time-series annotation. Export to all major ML formats.",
+      "categories": ["ai", "mlops", "data-labeling", "annotation"],
+      "platform": "linux",
+      "logo": "https://labelstud.io/images/ls-logo.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/label-studio/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "LS_PORT",
+          "label": "Web UI port",
+          "default": "8080"
+        },
+        {
+          "name": "LS_USER",
+          "label": "Admin email",
+          "default": "admin@example.com"
+        },
+        {
+          "name": "LS_PASSWORD",
+          "label": "Admin password",
+          "default": "changeme"
+        }
+      ]
+    },
+    {
+      "id": 17,
+      "type": 3,
+      "title": "Jupyter (GPU / PyTorch)",
+      "description": "GPU-accelerated Jupyter Lab with PyTorch, CUDA, and data science libraries pre-installed",
+      "note": "Requires NVIDIA GPU. Access with the configured token. Workspace persists in the work volume.",
+      "categories": ["ai", "ml-development", "notebooks", "pytorch"],
+      "platform": "linux",
+      "logo": "https://jupyter.org/assets/homepage/main-logo.svg",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/jupyter-gpu/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "JUPYTER_PORT",
+          "label": "Jupyter Lab port",
+          "default": "8888"
+        },
+        {
+          "name": "JUPYTER_TOKEN",
+          "label": "Access token",
+          "default": "changeme"
+        },
+        {
+          "name": "GRANT_SUDO",
+          "label": "Allow sudo in notebooks",
+          "default": "yes"
+        }
+      ]
+    },
+    {
+      "id": 18,
+      "type": 3,
+      "title": "Whisper ASR",
+      "description": "OpenAI Whisper speech-to-text API server with GPU acceleration. Supports transcription and translation",
+      "note": "Requires NVIDIA GPU. API documentation available at /docs. Supports models: tiny, base, small, medium, large-v3.",
+      "categories": ["ai", "speech-to-text", "transcription", "audio"],
+      "platform": "linux",
+      "logo": "https://upload.wikimedia.org/wikipedia/commons/0/04/ChatGPT_logo.svg",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/whisper/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "WHISPER_PORT",
+          "label": "API port",
+          "default": "9000"
+        },
+        {
+          "name": "ASR_MODEL",
+          "label": "Whisper model size",
+          "description": "Options: tiny, base, small, medium, large-v3",
+          "default": "base"
+        },
+        {
+          "name": "ASR_ENGINE",
+          "label": "ASR engine",
+          "default": "openai_whisper"
+        }
+      ]
+    },
+    {
+      "id": 19,
+      "type": 3,
+      "title": "NVIDIA Triton Inference Server",
+      "description": "Production-grade inference serving for any AI model — supports TensorRT, ONNX, PyTorch, TensorFlow, vLLM, and Python backends with dynamic batching, model ensembles, and multi-GPU scheduling",
+      "note": "Requires NVIDIA GPU. Place model repositories in the models volume following Triton's <a href=\"https://docs.nvidia.com/deeplearning/triton-inference-server/user-guide/docs/user_guide/model_repository.html\" target=\"_blank\">model repository layout</a>. Health check at /v2/health/ready.",
+      "categories": ["ai", "inference", "edge", "production", "nvidia"],
+      "platform": "linux",
+      "logo": "https://developer.nvidia.com/favicon.ico",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/triton/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "HTTP_PORT",
+          "label": "HTTP inference port",
+          "default": "8000"
+        },
+        {
+          "name": "GRPC_PORT",
+          "label": "gRPC inference port",
+          "default": "8001"
+        },
+        {
+          "name": "METRICS_PORT",
+          "label": "Prometheus metrics port",
+          "default": "8002"
+        },
+        {
+          "name": "TRITON_VERSION",
+          "label": "Triton version tag",
+          "default": "24.08"
+        },
+        {
+          "name": "MODEL_CONTROL",
+          "label": "Model control mode (none, poll, explicit)",
+          "default": "poll"
+        },
+        {
+          "name": "POLL_INTERVAL",
+          "label": "Model repository poll interval (seconds)",
+          "default": "30"
+        },
+        {
+          "name": "SHM_SIZE",
+          "label": "Shared memory size",
+          "default": "1g"
+        }
+      ]
+    },
+    {
+      "id": 20,
+      "type": 3,
+      "title": "ONNX Runtime Server",
+      "description": "Lightweight cross-platform inference server for ONNX models. Supports GPU and CPU-only profiles for edge deployment on resource-constrained nodes",
+      "note": "Use <code>docker compose --profile gpu up</code> for GPU nodes or <code>--profile edge up</code> for CPU-only edge nodes. Place your .onnx model file in the models volume.",
+      "categories": ["ai", "inference", "edge", "lightweight", "onnx"],
+      "platform": "linux",
+      "logo": "https://onnxruntime.ai/images/icons/ONNX-Runtime-logo.svg",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/onnx-runtime/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "HTTP_PORT",
+          "label": "HTTP port",
+          "default": "8001"
+        },
+        {
+          "name": "GRPC_PORT",
+          "label": "gRPC port",
+          "default": "50051"
+        },
+        {
+          "name": "MODEL_FILE",
+          "label": "Model filename in /models",
+          "default": "model.onnx"
+        },
+        {
+          "name": "NUM_THREADS",
+          "label": "Inference threads",
+          "default": "4"
+        },
+        {
+          "name": "CPU_LIMIT",
+          "label": "CPU core limit (edge profile)",
+          "default": "2.0"
+        },
+        {
+          "name": "MEM_LIMIT",
+          "label": "Memory limit (edge profile)",
+          "default": "2G"
+        }
+      ]
+    },
+    {
+      "id": 21,
+      "type": 3,
+      "title": "NVIDIA DeepStream",
+      "description": "GPU-accelerated video analytics and computer vision pipeline for industrial inspection, anomaly detection, and smart factory applications with Triton backend",
+      "note": "Requires NVIDIA GPU with video decode capabilities. For camera access on edge devices, set PRIVILEGED=true. Supports RTSP output on port 8554.",
+      "categories": ["ai", "computer-vision", "industrial", "edge", "video-analytics"],
+      "platform": "linux",
+      "logo": "https://developer.nvidia.com/favicon.ico",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/deepstream/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "RTSP_PORT",
+          "label": "RTSP output port",
+          "default": "8554"
+        },
+        {
+          "name": "REST_PORT",
+          "label": "REST API port",
+          "default": "9000"
+        },
+        {
+          "name": "DS_VERSION",
+          "label": "DeepStream version",
+          "default": "7.1"
+        },
+        {
+          "name": "SHM_SIZE",
+          "label": "Shared memory size",
+          "default": "2g"
+        },
+        {
+          "name": "PRIVILEGED",
+          "label": "Privileged mode (for device access)",
+          "default": "false"
+        }
+      ]
+    },
+    {
+      "id": 22,
+      "type": 3,
+      "title": "Ray Cluster (GPU)",
+      "description": "Distributed compute cluster for LLM fine-tuning, distributed training, hyperparameter tuning, and scalable inference with Ray Serve. Head + configurable worker nodes",
+      "note": "Requires NVIDIA GPU on all nodes. Scale workers with NUM_WORKERS. Dashboard accessible at the configured port. Includes Ray Train, Tune, Serve, and Data.",
+      "categories": ["ai", "distributed-training", "fine-tuning", "inference", "cluster"],
+      "platform": "linux",
+      "logo": "https://docs.ray.io/en/latest/_static/ray_logo.png",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/ray-cluster/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "DASHBOARD_PORT",
+          "label": "Ray Dashboard port",
+          "default": "8265"
+        },
+        {
+          "name": "SERVE_PORT",
+          "label": "Ray Serve port",
+          "default": "8000"
+        },
+        {
+          "name": "RAY_VERSION",
+          "label": "Ray version",
+          "default": "2.40.0"
+        },
+        {
+          "name": "NUM_WORKERS",
+          "label": "Number of worker nodes",
+          "default": "1"
+        },
+        {
+          "name": "HEAD_GPUS",
+          "label": "GPUs on head node",
+          "default": "1"
+        },
+        {
+          "name": "WORKER_GPUS",
+          "label": "GPUs per worker",
+          "default": "1"
+        },
+        {
+          "name": "WORKER_CPUS",
+          "label": "CPUs per worker",
+          "default": "4"
+        },
+        {
+          "name": "SHM_SIZE",
+          "label": "Shared memory per node",
+          "default": "8g"
+        }
+      ]
+    },
+    {
+      "id": 23,
+      "type": 3,
+      "title": "Prefect (ML Pipeline Orchestration)",
+      "description": "Governed ML pipeline orchestration platform with scheduling, retries, audit logging, and role-based access. Includes server, worker, and PostgreSQL backend",
+      "note": "Access the Prefect UI at the configured port. Create flows in Python and register them against this server. Worker uses Docker execution for isolation.",
+      "categories": ["ai", "mlops", "pipelines", "governance", "orchestration"],
+      "platform": "linux",
+      "logo": "https://www.prefect.io/favicon.ico",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/prefect/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "PREFECT_PORT",
+          "label": "Prefect UI port",
+          "default": "4200"
+        },
+        {
+          "name": "PREFECT_VERSION",
+          "label": "Prefect version",
+          "default": "3-latest"
+        },
+        {
+          "name": "PG_USER",
+          "label": "PostgreSQL user",
+          "default": "prefect"
+        },
+        {
+          "name": "PG_PASSWORD",
+          "label": "PostgreSQL password",
+          "default": "prefect"
+        },
+        {
+          "name": "ANALYTICS",
+          "label": "Enable analytics",
+          "default": "false"
+        }
+      ]
+    },
+    {
+      "id": 24,
+      "type": 3,
+      "title": "BentoML",
+      "description": "Unified model serving framework for packaging, deploying, and managing ML models as production-ready API endpoints with GPU support",
+      "note": "Requires NVIDIA GPU. Build Bentos (model packages) and serve them through this runtime. Prometheus metrics on port 3001.",
+      "categories": ["ai", "model-serving", "inference", "mlops"],
+      "platform": "linux",
+      "logo": "https://docs.bentoml.com/en/latest/_static/img/logo.svg",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/bentoml/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "BENTO_PORT",
+          "label": "Serving API port",
+          "default": "3000"
+        },
+        {
+          "name": "METRICS_PORT",
+          "label": "Prometheus metrics port",
+          "default": "3001"
+        },
+        {
+          "name": "BENTO_VERSION",
+          "label": "BentoML version",
+          "default": "latest"
+        },
+        {
+          "name": "LOG_LEVEL",
+          "label": "Log level",
+          "default": "INFO"
+        }
+      ]
+    },
+    {
+      "id": 25,
+      "type": 3,
+      "title": "MLflow + MinIO (Production MLOps)",
+      "description": "Production-grade MLOps stack: MLflow tracking server with PostgreSQL backend and MinIO S3-compatible artifact store for governed model registry, experiment tracking, and versioned artifact storage",
+      "note": "MinIO console available at port 9001. MLflow auto-creates the artifact bucket on startup. For production, change all default credentials.",
+      "categories": ["ai", "mlops", "model-registry", "governance", "experiment-tracking"],
+      "platform": "linux",
+      "logo": "https://mlflow.org/img/mlflow-black.svg",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/minio-mlops/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "MLFLOW_PORT",
+          "label": "MLflow UI port",
+          "default": "5000"
+        },
+        {
+          "name": "MINIO_API_PORT",
+          "label": "MinIO S3 API port",
+          "default": "9000"
+        },
+        {
+          "name": "MINIO_CONSOLE_PORT",
+          "label": "MinIO console port",
+          "default": "9001"
+        },
+        {
+          "name": "PG_USER",
+          "label": "PostgreSQL user",
+          "default": "mlflow"
+        },
+        {
+          "name": "PG_PASSWORD",
+          "label": "PostgreSQL password",
+          "default": "mlflow"
+        },
+        {
+          "name": "MINIO_ROOT_USER",
+          "label": "MinIO root user",
+          "default": "mlflow"
+        },
+        {
+          "name": "MINIO_ROOT_PASSWORD",
+          "label": "MinIO root password",
+          "default": "mlflow123"
+        },
+        {
+          "name": "ARTIFACT_BUCKET",
+          "label": "S3 artifact bucket name",
+          "default": "mlflow-artifacts"
+        }
+      ]
+    },
+    {
+      "id": 26,
+      "type": 3,
+      "title": "NVIDIA NIM",
+      "description": "Enterprise-grade optimized LLM inference microservice from NVIDIA. Pre-optimized with TensorRT-LLM for maximum throughput with OpenAI-compatible API",
+      "note": "Requires NVIDIA GPU and an NGC API key from <a href=\"https://build.nvidia.com/\" target=\"_blank\">NVIDIA Build</a>. Model downloads are cached in the nim_cache volume. First startup may take several minutes.",
+      "categories": ["ai", "llm", "inference", "enterprise", "nvidia"],
+      "platform": "linux",
+      "logo": "https://developer.nvidia.com/favicon.ico",
+      "repository": {
+        "url": "https://git.oe74.net/adelorenzo/portainer_scripts",
+        "stackfile": "ai-templates/stacks/nvidia-nim/docker-compose.yml"
+      },
+      "env": [
+        {
+          "name": "NIM_PORT",
+          "label": "API port",
+          "default": "8000"
+        },
+        {
+          "name": "NGC_API_KEY",
+          "label": "NVIDIA NGC API key (required)"
+        },
+        {
+          "name": "NIM_MODEL",
+          "label": "NIM model container",
+          "description": "Model from NVIDIA NGC catalog",
+          "default": "meta/llama-3.1-8b-instruct"
+        },
+        {
+          "name": "NIM_VERSION",
+          "label": "NIM version",
+          "default": "latest"
+        },
+        {
+          "name": "MAX_MODEL_LEN",
+          "label": "Max sequence length",
+          "default": "4096"
+        },
+        {
+          "name": "GPU_MEM_UTIL",
+          "label": "GPU memory utilization (0-1)",
+          "default": "0.9"
+        },
+        {
+          "name": "SHM_SIZE",
+          "label": "Shared memory size",
+          "default": "16g"
+        }
+      ]
+    }
+  ]
+}
diff --git a/ai-templates-0/stacks/bentoml/docker-compose.yml b/ai-templates-0/stacks/bentoml/docker-compose.yml
new file mode 100644
index 0000000..8833740
--- /dev/null
+++ b/ai-templates-0/stacks/bentoml/docker-compose.yml
@@ -0,0 +1,30 @@
+version: "3.8"
+
+services:
+  bentoml:
+    image: bentoml/bentoml:${BENTO_VERSION:-latest}
+    container_name: bentoml
+    restart: unless-stopped
+    ports:
+      - "${BENTO_PORT:-3000}:3000"
+      - "${METRICS_PORT:-3001}:3001"
+    volumes:
+      - bentoml_home:/home/bentoml
+      - bentoml_models:/home/bentoml/bentoml/models
+    environment:
+      - BENTOML_HOME=/home/bentoml/bentoml
+      - BENTOML_PORT=3000
+      - BENTOML_METRICS_PORT=3001
+      - BENTOML_LOG_LEVEL=${LOG_LEVEL:-INFO}
+    command: bentoml serve --host 0.0.0.0 --port 3000
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+
+volumes:
+  bentoml_home:
+  bentoml_models:
diff --git a/ai-templates-0/stacks/chromadb/docker-compose.yml b/ai-templates-0/stacks/chromadb/docker-compose.yml
new file mode 100644
index 0000000..f0a9496
--- /dev/null
+++ b/ai-templates-0/stacks/chromadb/docker-compose.yml
@@ -0,0 +1,20 @@
+version: "3.8"
+
+services:
+  chromadb:
+    image: chromadb/chroma:latest
+    container_name: chromadb
+    restart: unless-stopped
+    ports:
+      - "${CHROMA_PORT:-8000}:8000"
+    volumes:
+      - chroma_data:/chroma/chroma
+    environment:
+      - IS_PERSISTENT=TRUE
+      - PERSIST_DIRECTORY=/chroma/chroma
+      - ANONYMIZED_TELEMETRY=${TELEMETRY:-FALSE}
+      - CHROMA_SERVER_AUTHN_CREDENTIALS=${CHROMA_TOKEN:-}
+      - CHROMA_SERVER_AUTHN_PROVIDER=${CHROMA_AUTH_PROVIDER:-}
+
+volumes:
+  chroma_data:
diff --git a/ai-templates-0/stacks/comfyui/docker-compose.yml b/ai-templates-0/stacks/comfyui/docker-compose.yml
new file mode 100644
index 0000000..f25f4b1
--- /dev/null
+++ b/ai-templates-0/stacks/comfyui/docker-compose.yml
@@ -0,0 +1,31 @@
+version: "3.8"
+
+services:
+  comfyui:
+    image: yanwk/comfyui-boot:latest
+    container_name: comfyui
+    restart: unless-stopped
+    ports:
+      - "${COMFYUI_PORT:-8188}:8188"
+    volumes:
+      - comfyui_data:/root
+      - comfyui_models:/root/ComfyUI/models
+      - comfyui_output:/root/ComfyUI/output
+      - comfyui_input:/root/ComfyUI/input
+      - comfyui_custom_nodes:/root/ComfyUI/custom_nodes
+    environment:
+      - CLI_ARGS=${CLI_ARGS:---listen 0.0.0.0 --port 8188}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+
+volumes:
+  comfyui_data:
+  comfyui_models:
+  comfyui_output:
+  comfyui_input:
+  comfyui_custom_nodes:
diff --git a/ai-templates-0/stacks/deepstream/docker-compose.yml b/ai-templates-0/stacks/deepstream/docker-compose.yml
new file mode 100644
index 0000000..a338d72
--- /dev/null
+++ b/ai-templates-0/stacks/deepstream/docker-compose.yml
@@ -0,0 +1,38 @@
+version: "3.8"
+
+services:
+  deepstream:
+    image: nvcr.io/nvidia/deepstream:${DS_VERSION:-7.1}-triton-multiarch
+    container_name: deepstream
+    restart: unless-stopped
+    ports:
+      - "${RTSP_PORT:-8554}:8554"
+      - "${REST_PORT:-9000}:9000"
+    volumes:
+      - deepstream_apps:/opt/nvidia/deepstream/deepstream/sources/apps
+      - deepstream_models:/opt/nvidia/deepstream/deepstream/samples/models
+      - deepstream_configs:/opt/nvidia/deepstream/deepstream/samples/configs
+      - deepstream_streams:/opt/nvidia/deepstream/deepstream/samples/streams
+    environment:
+      - CUDA_VISIBLE_DEVICES=${CUDA_DEVICES:-all}
+      - DISPLAY=${DISPLAY:-}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu, video, compute, utility]
+    runtime: nvidia
+    network_mode: ${NETWORK_MODE:-bridge}
+    shm_size: ${SHM_SIZE:-2g}
+    # Required for video device access on edge nodes
+    privileged: ${PRIVILEGED:-false}
+    devices:
+      - /dev/video0:/dev/video0
+
+volumes:
+  deepstream_apps:
+  deepstream_models:
+  deepstream_configs:
+  deepstream_streams:
diff --git a/ai-templates-0/stacks/flowise/docker-compose.yml b/ai-templates-0/stacks/flowise/docker-compose.yml
new file mode 100644
index 0000000..d877614
--- /dev/null
+++ b/ai-templates-0/stacks/flowise/docker-compose.yml
@@ -0,0 +1,19 @@
+version: "3.8"
+
+services:
+  flowise:
+    image: flowiseai/flowise:latest
+    container_name: flowise
+    restart: unless-stopped
+    ports:
+      - "${FLOWISE_PORT:-3000}:3000"
+    volumes:
+      - flowise_data:/root/.flowise
+    environment:
+      - FLOWISE_USERNAME=${FLOWISE_USERNAME:-admin}
+      - FLOWISE_PASSWORD=${FLOWISE_PASSWORD:-changeme}
+      - APIKEY_PATH=/root/.flowise
+      - LOG_PATH=/root/.flowise/logs
+
+volumes:
+  flowise_data:
diff --git a/ai-templates-0/stacks/jupyter-gpu/docker-compose.yml b/ai-templates-0/stacks/jupyter-gpu/docker-compose.yml
new file mode 100644
index 0000000..b8554c4
--- /dev/null
+++ b/ai-templates-0/stacks/jupyter-gpu/docker-compose.yml
@@ -0,0 +1,26 @@
+version: "3.8"
+
+services:
+  jupyter:
+    image: quay.io/jupyter/pytorch-notebook:latest
+    container_name: jupyter-gpu
+    restart: unless-stopped
+    ports:
+      - "${JUPYTER_PORT:-8888}:8888"
+    volumes:
+      - jupyter_data:/home/jovyan/work
+    environment:
+      - JUPYTER_TOKEN=${JUPYTER_TOKEN:-changeme}
+      - JUPYTER_ENABLE_LAB=yes
+      - GRANT_SUDO=${GRANT_SUDO:-yes}
+    user: root
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+
+volumes:
+  jupyter_data:
diff --git a/ai-templates-0/stacks/label-studio/docker-compose.yml b/ai-templates-0/stacks/label-studio/docker-compose.yml
new file mode 100644
index 0000000..50558b7
--- /dev/null
+++ b/ai-templates-0/stacks/label-studio/docker-compose.yml
@@ -0,0 +1,19 @@
+version: "3.8"
+
+services:
+  label-studio:
+    image: heartexlabs/label-studio:latest
+    container_name: label-studio
+    restart: unless-stopped
+    ports:
+      - "${LS_PORT:-8080}:8080"
+    volumes:
+      - label_studio_data:/label-studio/data
+    environment:
+      - LABEL_STUDIO_LOCAL_FILES_SERVING_ENABLED=true
+      - LABEL_STUDIO_LOCAL_FILES_DOCUMENT_ROOT=/label-studio/data/files
+      - LABEL_STUDIO_USERNAME=${LS_USER:-admin@example.com}
+      - LABEL_STUDIO_PASSWORD=${LS_PASSWORD:-changeme}
+
+volumes:
+  label_studio_data:
diff --git a/ai-templates-0/stacks/langflow/docker-compose.yml b/ai-templates-0/stacks/langflow/docker-compose.yml
new file mode 100644
index 0000000..965b524
--- /dev/null
+++ b/ai-templates-0/stacks/langflow/docker-compose.yml
@@ -0,0 +1,18 @@
+version: "3.8"
+
+services:
+  langflow:
+    image: langflowai/langflow:latest
+    container_name: langflow
+    restart: unless-stopped
+    ports:
+      - "${LANGFLOW_PORT:-7860}:7860"
+    volumes:
+      - langflow_data:/app/langflow
+    environment:
+      - LANGFLOW_DATABASE_URL=sqlite:////app/langflow/langflow.db
+      - LANGFLOW_CONFIG_DIR=/app/langflow
+      - LANGFLOW_AUTO_LOGIN=${AUTO_LOGIN:-true}
+
+volumes:
+  langflow_data:
diff --git a/ai-templates-0/stacks/litellm/docker-compose.yml b/ai-templates-0/stacks/litellm/docker-compose.yml
new file mode 100644
index 0000000..e381fbb
--- /dev/null
+++ b/ai-templates-0/stacks/litellm/docker-compose.yml
@@ -0,0 +1,33 @@
+version: "3.8"
+
+services:
+  litellm:
+    image: ghcr.io/berriai/litellm:main-latest
+    container_name: litellm
+    restart: unless-stopped
+    ports:
+      - "${LITELLM_PORT:-4000}:4000"
+    volumes:
+      - litellm_config:/app/config
+    environment:
+      - LITELLM_MASTER_KEY=${LITELLM_MASTER_KEY:-sk-master-key}
+      - LITELLM_LOG_LEVEL=${LOG_LEVEL:-INFO}
+      - DATABASE_URL=postgresql://${PG_USER:-litellm}:${PG_PASSWORD:-litellm}@litellm-db:5432/${PG_DB:-litellm}
+    command: --config /app/config/litellm_config.yaml --port 4000
+    depends_on:
+      - litellm-db
+
+  litellm-db:
+    image: postgres:16-alpine
+    container_name: litellm-db
+    restart: unless-stopped
+    environment:
+      - POSTGRES_USER=${PG_USER:-litellm}
+      - POSTGRES_PASSWORD=${PG_PASSWORD:-litellm}
+      - POSTGRES_DB=${PG_DB:-litellm}
+    volumes:
+      - litellm_pg_data:/var/lib/postgresql/data
+
+volumes:
+  litellm_config:
+  litellm_pg_data:
diff --git a/ai-templates-0/stacks/localai/docker-compose.yml b/ai-templates-0/stacks/localai/docker-compose.yml
new file mode 100644
index 0000000..9c5f8fe
--- /dev/null
+++ b/ai-templates-0/stacks/localai/docker-compose.yml
@@ -0,0 +1,25 @@
+version: "3.8"
+
+services:
+  localai:
+    image: localai/localai:latest-gpu-nvidia-cuda-12
+    container_name: localai
+    restart: unless-stopped
+    ports:
+      - "${LOCALAI_PORT:-8080}:8080"
+    volumes:
+      - localai_models:/build/models
+    environment:
+      - THREADS=${THREADS:-4}
+      - CONTEXT_SIZE=${CONTEXT_SIZE:-4096}
+      - MODELS_PATH=/build/models
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+
+volumes:
+  localai_models:
diff --git a/ai-templates-0/stacks/minio-mlops/docker-compose.yml b/ai-templates-0/stacks/minio-mlops/docker-compose.yml
new file mode 100644
index 0000000..a682d08
--- /dev/null
+++ b/ai-templates-0/stacks/minio-mlops/docker-compose.yml
@@ -0,0 +1,76 @@
+version: "3.8"
+
+services:
+  mlflow:
+    image: ghcr.io/mlflow/mlflow:${MLFLOW_VERSION:-latest}
+    container_name: mlflow-server
+    restart: unless-stopped
+    ports:
+      - "${MLFLOW_PORT:-5000}:5000"
+    environment:
+      - MLFLOW_TRACKING_URI=postgresql://${PG_USER:-mlflow}:${PG_PASSWORD:-mlflow}@mlflow-db:5432/${PG_DB:-mlflow}
+      - MLFLOW_S3_ENDPOINT_URL=http://mlflow-minio:9000
+      - AWS_ACCESS_KEY_ID=${MINIO_ROOT_USER:-mlflow}
+      - AWS_SECRET_ACCESS_KEY=${MINIO_ROOT_PASSWORD:-mlflow123}
+      - MLFLOW_DEFAULT_ARTIFACT_ROOT=s3://${ARTIFACT_BUCKET:-mlflow-artifacts}/
+    command: >
+      mlflow server
+      --host 0.0.0.0
+      --port 5000
+      --backend-store-uri postgresql://${PG_USER:-mlflow}:${PG_PASSWORD:-mlflow}@mlflow-db:5432/${PG_DB:-mlflow}
+      --default-artifact-root s3://${ARTIFACT_BUCKET:-mlflow-artifacts}/
+      --serve-artifacts
+    depends_on:
+      mlflow-db:
+        condition: service_healthy
+      mlflow-minio:
+        condition: service_started
+
+  mlflow-db:
+    image: postgres:16-alpine
+    container_name: mlflow-db
+    restart: unless-stopped
+    environment:
+      - POSTGRES_USER=${PG_USER:-mlflow}
+      - POSTGRES_PASSWORD=${PG_PASSWORD:-mlflow}
+      - POSTGRES_DB=${PG_DB:-mlflow}
+    volumes:
+      - mlflow_pg_data:/var/lib/postgresql/data
+    healthcheck:
+      test: ["CMD-SHELL", "pg_isready -U ${PG_USER:-mlflow}"]
+      interval: 10s
+      timeout: 5s
+      retries: 5
+
+  mlflow-minio:
+    image: quay.io/minio/minio:latest
+    container_name: mlflow-minio
+    restart: unless-stopped
+    ports:
+      - "${MINIO_API_PORT:-9000}:9000"
+      - "${MINIO_CONSOLE_PORT:-9001}:9001"
+    volumes:
+      - mlflow_minio_data:/data
+    environment:
+      - MINIO_ROOT_USER=${MINIO_ROOT_USER:-mlflow}
+      - MINIO_ROOT_PASSWORD=${MINIO_ROOT_PASSWORD:-mlflow123}
+    command: server /data --console-address ':9001'
+
+  # Init container to create the default bucket
+  mlflow-minio-init:
+    image: quay.io/minio/mc:latest
+    container_name: mlflow-minio-init
+    depends_on:
+      - mlflow-minio
+    entrypoint: >
+      /bin/sh -c "
+      sleep 5;
+      mc alias set myminio http://mlflow-minio:9000 ${MINIO_ROOT_USER:-mlflow} ${MINIO_ROOT_PASSWORD:-mlflow123};
+      mc mb --ignore-existing myminio/${ARTIFACT_BUCKET:-mlflow-artifacts};
+      mc anonymous set download myminio/${ARTIFACT_BUCKET:-mlflow-artifacts};
+      exit 0;
+      "
+
+volumes:
+  mlflow_pg_data:
+  mlflow_minio_data:
diff --git a/ai-templates-0/stacks/mlflow/docker-compose.yml b/ai-templates-0/stacks/mlflow/docker-compose.yml
new file mode 100644
index 0000000..958504b
--- /dev/null
+++ b/ai-templates-0/stacks/mlflow/docker-compose.yml
@@ -0,0 +1,20 @@
+version: "3.8"
+
+services:
+  mlflow:
+    image: ghcr.io/mlflow/mlflow:latest
+    container_name: mlflow
+    restart: unless-stopped
+    ports:
+      - "${MLFLOW_PORT:-5000}:5000"
+    volumes:
+      - mlflow_data:/mlflow
+    command: >
+      mlflow server
+      --host 0.0.0.0
+      --port 5000
+      --backend-store-uri sqlite:///mlflow/mlflow.db
+      --default-artifact-root /mlflow/artifacts
+
+volumes:
+  mlflow_data:
diff --git a/ai-templates-0/stacks/n8n-ai/docker-compose.yml b/ai-templates-0/stacks/n8n-ai/docker-compose.yml
new file mode 100644
index 0000000..30ac80e
--- /dev/null
+++ b/ai-templates-0/stacks/n8n-ai/docker-compose.yml
@@ -0,0 +1,20 @@
+version: "3.8"
+
+services:
+  n8n:
+    image: docker.n8n.io/n8nio/n8n:latest
+    container_name: n8n-ai
+    restart: unless-stopped
+    ports:
+      - "${N8N_PORT:-5678}:5678"
+    volumes:
+      - n8n_data:/home/node/.n8n
+    environment:
+      - N8N_BASIC_AUTH_ACTIVE=${N8N_AUTH:-true}
+      - N8N_BASIC_AUTH_USER=${N8N_USER:-admin}
+      - N8N_BASIC_AUTH_PASSWORD=${N8N_PASSWORD:-changeme}
+      - WEBHOOK_URL=${WEBHOOK_URL:-http://localhost:5678/}
+      - N8N_AI_ENABLED=true
+
+volumes:
+  n8n_data:
diff --git a/ai-templates-0/stacks/nvidia-nim/docker-compose.yml b/ai-templates-0/stacks/nvidia-nim/docker-compose.yml
new file mode 100644
index 0000000..77976bd
--- /dev/null
+++ b/ai-templates-0/stacks/nvidia-nim/docker-compose.yml
@@ -0,0 +1,36 @@
+version: "3.8"
+
+services:
+  nim:
+    image: nvcr.io/nim/${NIM_MODEL:-meta/llama-3.1-8b-instruct}:${NIM_VERSION:-latest}
+    container_name: nvidia-nim
+    restart: unless-stopped
+    ports:
+      - "${NIM_PORT:-8000}:8000"
+    volumes:
+      - nim_cache:/opt/nim/.cache
+    environment:
+      - NGC_API_KEY=${NGC_API_KEY}
+      - NIM_MAX_MODEL_LEN=${MAX_MODEL_LEN:-4096}
+      - NIM_GPU_MEMORY_UTILIZATION=${GPU_MEM_UTIL:-0.9}
+      - NIM_MAX_BATCH_SIZE=${MAX_BATCH:-256}
+      - NIM_LOG_LEVEL=${LOG_LEVEL:-INFO}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+    shm_size: ${SHM_SIZE:-16g}
+    ulimits:
+      memlock: -1
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://localhost:8000/v1/health/ready"]
+      interval: 30s
+      timeout: 10s
+      retries: 10
+      start_period: 120s
+
+volumes:
+  nim_cache:
diff --git a/ai-templates-0/stacks/ollama/docker-compose.yml b/ai-templates-0/stacks/ollama/docker-compose.yml
new file mode 100644
index 0000000..a10b813
--- /dev/null
+++ b/ai-templates-0/stacks/ollama/docker-compose.yml
@@ -0,0 +1,25 @@
+version: "3.8"
+
+services:
+  ollama:
+    image: ollama/ollama:latest
+    container_name: ollama
+    restart: unless-stopped
+    ports:
+      - "${OLLAMA_PORT:-11434}:11434"
+    volumes:
+      - ollama_data:/root/.ollama
+    environment:
+      - OLLAMA_HOST=0.0.0.0
+      - OLLAMA_NUM_PARALLEL=${OLLAMA_NUM_PARALLEL:-4}
+      - OLLAMA_MAX_LOADED_MODELS=${OLLAMA_MAX_LOADED_MODELS:-2}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+
+volumes:
+  ollama_data:
diff --git a/ai-templates-0/stacks/onnx-runtime/docker-compose.yml b/ai-templates-0/stacks/onnx-runtime/docker-compose.yml
new file mode 100644
index 0000000..16ff061
--- /dev/null
+++ b/ai-templates-0/stacks/onnx-runtime/docker-compose.yml
@@ -0,0 +1,57 @@
+version: "3.8"
+
+services:
+  # GPU variant — for data center / cloud nodes
+  onnx-runtime-gpu:
+    image: mcr.microsoft.com/onnxruntime/server:latest
+    container_name: onnx-runtime-gpu
+    restart: unless-stopped
+    profiles: ["gpu"]
+    ports:
+      - "${HTTP_PORT:-8001}:8001"
+      - "${GRPC_PORT:-50051}:50051"
+    volumes:
+      - onnx_models:/models
+    environment:
+      - ORT_LOG_LEVEL=${LOG_LEVEL:-WARNING}
+    command: >
+      --model_path /models/${MODEL_FILE:-model.onnx}
+      --http_port 8001
+      --grpc_port 50051
+      --num_threads ${NUM_THREADS:-4}
+      --execution_provider ${EXEC_PROVIDER:-cuda}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: 1
+              capabilities: [gpu]
+
+  # CPU variant — for edge nodes, ARM, resource-constrained environments
+  onnx-runtime-cpu:
+    image: mcr.microsoft.com/onnxruntime/server:latest
+    container_name: onnx-runtime-cpu
+    restart: unless-stopped
+    profiles: ["cpu", "edge"]
+    ports:
+      - "${HTTP_PORT:-8001}:8001"
+      - "${GRPC_PORT:-50051}:50051"
+    volumes:
+      - onnx_models:/models
+    environment:
+      - ORT_LOG_LEVEL=${LOG_LEVEL:-WARNING}
+    command: >
+      --model_path /models/${MODEL_FILE:-model.onnx}
+      --http_port 8001
+      --grpc_port 50051
+      --num_threads ${NUM_THREADS:-4}
+      --execution_provider cpu
+    deploy:
+      resources:
+        limits:
+          cpus: "${CPU_LIMIT:-2.0}"
+          memory: ${MEM_LIMIT:-2G}
+
+volumes:
+  onnx_models:
diff --git a/ai-templates-0/stacks/open-webui/docker-compose.yml b/ai-templates-0/stacks/open-webui/docker-compose.yml
new file mode 100644
index 0000000..709f69f
--- /dev/null
+++ b/ai-templates-0/stacks/open-webui/docker-compose.yml
@@ -0,0 +1,39 @@
+version: "3.8"
+
+services:
+  open-webui:
+    image: ghcr.io/open-webui/open-webui:main
+    container_name: open-webui
+    restart: unless-stopped
+    ports:
+      - "${OPEN_WEBUI_PORT:-3000}:8080"
+    volumes:
+      - open_webui_data:/app/backend/data
+    environment:
+      - OLLAMA_BASE_URL=${OLLAMA_BASE_URL:-http://ollama:11434}
+      - WEBUI_SECRET_KEY=${WEBUI_SECRET_KEY:-changeme}
+      - ENABLE_SIGNUP=${ENABLE_SIGNUP:-true}
+    depends_on:
+      - ollama
+
+  ollama:
+    image: ollama/ollama:latest
+    container_name: ollama
+    restart: unless-stopped
+    ports:
+      - "${OLLAMA_PORT:-11434}:11434"
+    volumes:
+      - ollama_data:/root/.ollama
+    environment:
+      - OLLAMA_HOST=0.0.0.0
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+
+volumes:
+  open_webui_data:
+  ollama_data:
diff --git a/ai-templates-0/stacks/prefect/docker-compose.yml b/ai-templates-0/stacks/prefect/docker-compose.yml
new file mode 100644
index 0000000..c26748f
--- /dev/null
+++ b/ai-templates-0/stacks/prefect/docker-compose.yml
@@ -0,0 +1,55 @@
+version: "3.8"
+
+services:
+  prefect-server:
+    image: prefecthq/prefect:${PREFECT_VERSION:-3-latest}
+    container_name: prefect-server
+    restart: unless-stopped
+    ports:
+      - "${PREFECT_PORT:-4200}:4200"
+    volumes:
+      - prefect_data:/root/.prefect
+      - prefect_flows:/flows
+    environment:
+      - PREFECT_SERVER_API_HOST=0.0.0.0
+      - PREFECT_SERVER_API_PORT=4200
+      - PREFECT_API_DATABASE_CONNECTION_URL=postgresql+asyncpg://${PG_USER:-prefect}:${PG_PASSWORD:-prefect}@prefect-db:5432/${PG_DB:-prefect}
+      - PREFECT_SERVER_ANALYTICS_ENABLED=${ANALYTICS:-false}
+    command: prefect server start
+    depends_on:
+      prefect-db:
+        condition: service_healthy
+
+  prefect-worker:
+    image: prefecthq/prefect:${PREFECT_VERSION:-3-latest}
+    container_name: prefect-worker
+    restart: unless-stopped
+    volumes:
+      - prefect_flows:/flows
+      - /var/run/docker.sock:/var/run/docker.sock
+    environment:
+      - PREFECT_API_URL=http://prefect-server:4200/api
+    command: prefect worker start --pool default-agent-pool --type docker
+    depends_on:
+      - prefect-server
+
+  prefect-db:
+    image: postgres:16-alpine
+    container_name: prefect-db
+    restart: unless-stopped
+    environment:
+      - POSTGRES_USER=${PG_USER:-prefect}
+      - POSTGRES_PASSWORD=${PG_PASSWORD:-prefect}
+      - POSTGRES_DB=${PG_DB:-prefect}
+    volumes:
+      - prefect_pg_data:/var/lib/postgresql/data
+    healthcheck:
+      test: ["CMD-SHELL", "pg_isready -U ${PG_USER:-prefect}"]
+      interval: 10s
+      timeout: 5s
+      retries: 5
+
+volumes:
+  prefect_data:
+  prefect_flows:
+  prefect_pg_data:
diff --git a/ai-templates-0/stacks/qdrant/docker-compose.yml b/ai-templates-0/stacks/qdrant/docker-compose.yml
new file mode 100644
index 0000000..03ca69c
--- /dev/null
+++ b/ai-templates-0/stacks/qdrant/docker-compose.yml
@@ -0,0 +1,19 @@
+version: "3.8"
+
+services:
+  qdrant:
+    image: qdrant/qdrant:latest
+    container_name: qdrant
+    restart: unless-stopped
+    ports:
+      - "${QDRANT_HTTP_PORT:-6333}:6333"
+      - "${QDRANT_GRPC_PORT:-6334}:6334"
+    volumes:
+      - qdrant_data:/qdrant/storage
+      - qdrant_snapshots:/qdrant/snapshots
+    environment:
+      - QDRANT__SERVICE__API_KEY=${QDRANT_API_KEY:-}
+
+volumes:
+  qdrant_data:
+  qdrant_snapshots:
diff --git a/ai-templates-0/stacks/ray-cluster/docker-compose.yml b/ai-templates-0/stacks/ray-cluster/docker-compose.yml
new file mode 100644
index 0000000..be66943
--- /dev/null
+++ b/ai-templates-0/stacks/ray-cluster/docker-compose.yml
@@ -0,0 +1,60 @@
+version: "3.8"
+
+services:
+  ray-head:
+    image: rayproject/ray-ml:${RAY_VERSION:-2.40.0}-py310-gpu
+    container_name: ray-head
+    restart: unless-stopped
+    ports:
+      - "${DASHBOARD_PORT:-8265}:8265"
+      - "${CLIENT_PORT:-10001}:10001"
+      - "${GCS_PORT:-6379}:6379"
+      - "${SERVE_PORT:-8000}:8000"
+    volumes:
+      - ray_data:/home/ray/data
+      - ray_results:/home/ray/ray_results
+    command: >
+      ray start --head
+      --port=6379
+      --dashboard-host=0.0.0.0
+      --dashboard-port=8265
+      --num-gpus=${HEAD_GPUS:-1}
+      --block
+    environment:
+      - RAY_GRAFANA_HOST=http://grafana:3000
+      - RAY_PROMETHEUS_HOST=http://prometheus:9090
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+    shm_size: ${SHM_SIZE:-8g}
+
+  ray-worker:
+    image: rayproject/ray-ml:${RAY_VERSION:-2.40.0}-py310-gpu
+    restart: unless-stopped
+    depends_on:
+      - ray-head
+    command: >
+      ray start
+      --address=ray-head:6379
+      --num-gpus=${WORKER_GPUS:-1}
+      --num-cpus=${WORKER_CPUS:-4}
+      --block
+    volumes:
+      - ray_data:/home/ray/data
+    deploy:
+      replicas: ${NUM_WORKERS:-1}
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+    shm_size: ${SHM_SIZE:-8g}
+
+volumes:
+  ray_data:
+  ray_results:
diff --git a/ai-templates-0/stacks/stable-diffusion-webui/docker-compose.yml b/ai-templates-0/stacks/stable-diffusion-webui/docker-compose.yml
new file mode 100644
index 0000000..b4592de
--- /dev/null
+++ b/ai-templates-0/stacks/stable-diffusion-webui/docker-compose.yml
@@ -0,0 +1,25 @@
+version: "3.8"
+
+services:
+  stable-diffusion-webui:
+    image: universonic/stable-diffusion-webui:latest
+    container_name: stable-diffusion-webui
+    restart: unless-stopped
+    ports:
+      - "${SD_PORT:-7860}:7860"
+    volumes:
+      - sd_data:/data
+      - sd_output:/output
+    environment:
+      - CLI_ARGS=${CLI_ARGS:---listen --api --xformers}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+
+volumes:
+  sd_data:
+  sd_output:
diff --git a/ai-templates-0/stacks/text-generation-webui/docker-compose.yml b/ai-templates-0/stacks/text-generation-webui/docker-compose.yml
new file mode 100644
index 0000000..cb65de8
--- /dev/null
+++ b/ai-templates-0/stacks/text-generation-webui/docker-compose.yml
@@ -0,0 +1,35 @@
+version: "3.8"
+
+services:
+  text-gen-webui:
+    image: atinoda/text-generation-webui:default-nvidia
+    container_name: text-generation-webui
+    restart: unless-stopped
+    ports:
+      - "${WEBUI_PORT:-7860}:7860"
+      - "${API_PORT:-5000}:5000"
+      - "${STREAM_PORT:-5005}:5005"
+    volumes:
+      - tgw_characters:/app/characters
+      - tgw_loras:/app/loras
+      - tgw_models:/app/models
+      - tgw_presets:/app/presets
+      - tgw_prompts:/app/prompts
+      - tgw_extensions:/app/extensions
+    environment:
+      - EXTRA_LAUNCH_ARGS=${EXTRA_LAUNCH_ARGS:---listen --api}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+
+volumes:
+  tgw_characters:
+  tgw_loras:
+  tgw_models:
+  tgw_presets:
+  tgw_prompts:
+  tgw_extensions:
diff --git a/ai-templates-0/stacks/triton/docker-compose.yml b/ai-templates-0/stacks/triton/docker-compose.yml
new file mode 100644
index 0000000..39c4b56
--- /dev/null
+++ b/ai-templates-0/stacks/triton/docker-compose.yml
@@ -0,0 +1,43 @@
+version: "3.8"
+
+services:
+  triton:
+    image: nvcr.io/nvidia/tritonserver:${TRITON_VERSION:-24.08}-py3
+    container_name: triton-inference-server
+    restart: unless-stopped
+    ports:
+      - "${HTTP_PORT:-8000}:8000"
+      - "${GRPC_PORT:-8001}:8001"
+      - "${METRICS_PORT:-8002}:8002"
+    volumes:
+      - triton_models:/models
+    command: >
+      tritonserver
+      --model-repository=/models
+      --strict-model-config=${STRICT_CONFIG:-false}
+      --log-verbose=${LOG_VERBOSE:-0}
+      --exit-on-error=${EXIT_ON_ERROR:-false}
+      --rate-limit=${RATE_LIMIT:-off}
+      --model-control-mode=${MODEL_CONTROL:-poll}
+      --repository-poll-secs=${POLL_INTERVAL:-30}
+    environment:
+      - CUDA_VISIBLE_DEVICES=${CUDA_DEVICES:-all}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+    shm_size: ${SHM_SIZE:-1g}
+    ulimits:
+      memlock: -1
+      stack: 67108864
+    healthcheck:
+      test: ["CMD", "curl", "-f", "http://localhost:8000/v2/health/ready"]
+      interval: 30s
+      timeout: 10s
+      retries: 5
+
+volumes:
+  triton_models:
diff --git a/ai-templates-0/stacks/vllm/docker-compose.yml b/ai-templates-0/stacks/vllm/docker-compose.yml
new file mode 100644
index 0000000..4b8c06d
--- /dev/null
+++ b/ai-templates-0/stacks/vllm/docker-compose.yml
@@ -0,0 +1,29 @@
+version: "3.8"
+
+services:
+  vllm:
+    image: vllm/vllm-openai:latest
+    container_name: vllm
+    restart: unless-stopped
+    ports:
+      - "${VLLM_PORT:-8000}:8000"
+    volumes:
+      - vllm_cache:/root/.cache/huggingface
+    environment:
+      - HUGGING_FACE_HUB_TOKEN=${HF_TOKEN:-}
+    command: >
+      --model ${MODEL_NAME:-meta-llama/Llama-3.1-8B-Instruct}
+      --max-model-len ${MAX_MODEL_LEN:-4096}
+      --gpu-memory-utilization ${GPU_MEM_UTIL:-0.90}
+      --tensor-parallel-size ${TENSOR_PARALLEL:-1}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]
+    ipc: host
+
+volumes:
+  vllm_cache:
diff --git a/ai-templates-0/stacks/weaviate/docker-compose.yml b/ai-templates-0/stacks/weaviate/docker-compose.yml
new file mode 100644
index 0000000..df41b85
--- /dev/null
+++ b/ai-templates-0/stacks/weaviate/docker-compose.yml
@@ -0,0 +1,22 @@
+version: "3.8"
+
+services:
+  weaviate:
+    image: cr.weaviate.io/semitechnologies/weaviate:latest
+    container_name: weaviate
+    restart: unless-stopped
+    ports:
+      - "${WEAVIATE_HTTP_PORT:-8080}:8080"
+      - "${WEAVIATE_GRPC_PORT:-50051}:50051"
+    volumes:
+      - weaviate_data:/var/lib/weaviate
+    environment:
+      - QUERY_DEFAULTS_LIMIT=25
+      - AUTHENTICATION_ANONYMOUS_ACCESS_ENABLED=${ANON_ACCESS:-true}
+      - PERSISTENCE_DATA_PATH=/var/lib/weaviate
+      - DEFAULT_VECTORIZER_MODULE=${VECTORIZER:-none}
+      - CLUSTER_HOSTNAME=node1
+      - ENABLE_MODULES=${MODULES:-text2vec-transformers,generative-openai}
+
+volumes:
+  weaviate_data:
diff --git a/ai-templates-0/stacks/whisper/docker-compose.yml b/ai-templates-0/stacks/whisper/docker-compose.yml
new file mode 100644
index 0000000..6a04b81
--- /dev/null
+++ b/ai-templates-0/stacks/whisper/docker-compose.yml
@@ -0,0 +1,19 @@
+version: "3.8"
+
+services:
+  whisper:
+    image: onerahmet/openai-whisper-asr-webservice:latest-gpu
+    container_name: whisper-asr
+    restart: unless-stopped
+    ports:
+      - "${WHISPER_PORT:-9000}:9000"
+    environment:
+      - ASR_MODEL=${ASR_MODEL:-base}
+      - ASR_ENGINE=${ASR_ENGINE:-openai_whisper}
+    deploy:
+      resources:
+        reservations:
+          devices:
+            - driver: nvidia
+              count: all
+              capabilities: [gpu]