A production-grade autonomous AI engineering platform with multi-agent architecture, evolutionary optimization, and distributed execution.

• Multi-Agent System: Planner, Executor, Evaluator, Fixer agents
• Competitive Parallel Execution: Multiple candidates per step with Ray
• Self-Healing Loop: Automatic error detection and fixing
• Vector Memory: FAISS-based RAG for solution retrieval
• Evolutionary Optimization: Prompt evolution based on success rates
• Sandboxed Execution: Safe subprocess-based code execution

# Install dependencies
pip install -r requirements.txt

# Run with fallback LLM (no API key needed for testing)
export USE_FALLBACK_LLM=true
python -m backend.main