AI / ML2024-11
Full-Stack Multi-Agent
I wanted to see how far I could push autonomous AI — could a crew of agents handle a full project without me touching it? I built a system with 6 specialized agents (research, code, review) orchestrated through CrewAI and LangGraph. Each agent has its own role, tools, and memory.
Key Highlights
- They collaborate, pass context, and self-correct
- By parallelizing the workflows, total project execution time dropped by 80%
- It's my sandbox for testing what agentic AI can actually do end-to-end
Technology Stack
AI
CrewAILangGraphClaude 3.5Agents
Core
Python