DeFi Research Agent: From Data to Alpha

Most DeFi builders are drowning in data and still making decisions in the dark. The gap between raw onchain information and actionable alpha is not a data problem - it is an infrastructure problem. This workshop shows you how to close it. Son Dang, Backend Engineer at Birdeye, walks you through building a production-ready DeFi research agent system from scratch, powered by Birdeye's real-time market data. Birdeye Data is the infrastructure behind teams like Phantom, Raydium, Coinbase, and Bybit, delivering real-time, comprehensive data across tokens, trades, wallets, and markets on Solana and beyond. This is a live, hands-on workshop. You will go from raw on-chain data to a working multi-agent research pipeline in one session. Son covers how to wire the Birdeye MCP Server into a multi-agent system, how to build a configurable research pipeline that works with any model, and how to apply the output to real DeFi use cases. This is directly relevant to your Colosseum submission if you are building anything in DeFi, trading, AI agents, or analytics. A working data layer is the difference between a demo that impresses judges and a product that impresses users. If your project needs on-chain intelligence to function, this session gives you the fastest path to building it. What you will walk away with: a working DeFi research agent connected to live Birdeye data, a clear architecture for multi-agent pipelines you can extend for your own project, and an understanding of how production teams at Phantom and Raydium use on-chain data infrastructure to maintain their edge. Come with your laptop. Leave with a working agent.Can't make it IRL? We gotchu.Dial-in: meet.google.com/afb-gcqi-ita