Hi, I'm Chidi 👋
I built this to share
what I've learned.
I've been working with GenAI for the past years, both professionally for fun. TalkingCode is a project I built to share some of the state-of-the-art techniques I've picked up along the way, and to give you a chance to get to know me through the code I write.
Suggested starting points
Learning
In Public.
This agent is a sandbox for the techniques I find most interesting: multi-step reasoning, precise tool calling, and structural context. I'm sharing them here so we can see how they perform on real-world data.
Reasoning over Data
Using my GitHub code as a rich dataset to demonstrate how agents can navigate and interpret complex information.
Multi-Step Planning
The core of the project: agentic loops that plan, research, and synthesize answers through iterative reasoning.
Driven by Research
Techniques & Patterns
How It Works
The Process Behind the Agent
Understanding
We index repositories by preserving the structural hierarchy, making it easier for the AI to "see" the code.
Reasoning
Instead of a direct search, the agent takes a moment to plan which parts of the code are most relevant to your question.
Sharing
The final answer includes citations, so you can see exactly which files and logic the agent used to reach its conclusion.
Let's talk code.
I’m always happy to talk about architecture, patterns, or the techniques used here. Feel free to explore and see how it works.
Start a Conversation