Sentinel AI

An ultra-premium, production-grade Artificial Intelligence dashboard for Multimodal Active Monitoring, Video Intelligence, and Semantic RAG operations.

Built with Next.js, FastAPI, Celery, and Ollama.

![Active Sentinel Live Tracker](/Users/phani.m/.gemini/antigravity/brain/d3111712-5c2a-47f4-ba03-115bf8233b8b/sentinel_live_1774083070226.png)
<!-- slide -->
![Multimodal Analysis Sandbox](/Users/phani.m/.gemini/antigravity/brain/d3111712-5c2a-47f4-ba03-115bf8233b8b/nextjs_image_final_1774080964697.png)
<!-- slide -->
![Monitoring Telemetry](/Users/phani.m/.gemini/antigravity/brain/d3111712-5c2a-47f4-ba03-115bf8233b8b/final_monitoring_1774007479645.png)

⚡ Core Capabilities

• Active WebRTC Sentinel: Hook directly into physical webcams. Monitor streams live with bi-directional WebSockets and configure natural-language "Tripwires" to automatically map and raise threat events to the Incident Dashboard.
• Multimodal Sandbox: Run complex multimodal LLMs locally without cloud API costs using our SSE Fetch API chunk sequence decoders for lightning-fast DOM rendering.
• Video Intel Pipeline: Extract and chunk massive .mp4 payloads synchronously utilizing Celery + Redis worker detachments.
• Local RAG Integration: Semantic search against PDF documents with continuous all-MiniLM-L6-v2 embeddings in ChromaDB collections.
• Google Labs UI: A heavily elevated enterprise interface styled using Framer Motion micro-animations and a custom Terracotta/Desert Sand configuration.

🚀 Quickstart

Everything you need to launch the pipeline out-of-the-box.

# 1. Clone & Install
git clone https://github.com/Phani3108/Sentinel-AI.git
cd Sentinel-AI
pip install -r requirements.txt
npm install --prefix web

# 2. Boot Local Infrastructure (requires Redis/Celery)
docker-compose -f docker/docker-compose.yml up -d redis worker

# 3. Boot Web Sockets & FastAPI
uvicorn api.main:app --port 8080

# 4. Launch the Next.js UI (Google Labs Theme)
cd web && npm run dev
# Go to http://localhost:3000