SARANG is a high-fidelity, autonomous research environment designed to deconstruct complex scientific papers, extract formal mathematical frameworks, and synthesize reproducible implementations.
Powered by Gemini 2.0 Flash, SARANG orchestrates a specialized team of autonomous agents:
- Research Intelligence: Synthesizes high-level strategy and hypothesis deconstruction.
- Math Architect: Extracts and formalizes LaTeX equations and theorems.
- Implementation Specialist: Generates validated Python simulations and scripts.
- Visual Insights: Analyzes genomics data, figures, and visual evidence.
- Reproducibility Engineer: Ensures every claim is backed by executable validation.
- Peer Reviewer: Critiques logic and ensures scientific rigor.
- Compute Monitor: Optimizes resource allocation and cost efficiency.
- Backend Swarm Engine (Python): Pure Python FastAPI service managing WebSockets, multi-agent reasoning chains, tiered LLM routing (Gemini → Groq → Mock), and in-memory session management.
- Frontend Dashboard (Next.js): React UI with real-time agentic interactions.
# 1. Install Python & Node dependencies
pip install -r requirements.txt
cd dashboard && npm install && cd ..
# 2. Configure API keys
# Add your GEMINI_API_KEY and GROQ_API_KEY (comma separated for multiple keys)
cp .env.example .env
# 3. Launch everything
./dev.shOr manually:
# Terminal 1: Python agents (Port 8000)
source venv/bin/activate && PYTHONPATH=. python3 agents_service/api/main.py
# Terminal 2: Dashboard (Port 3000)
cd dashboard && npm run devOpen http://localhost:3000 in your browser.
SARANG — Scientific Analysis & Reproducible Agentic Network Gateway