Thoughts on AI systems, data architecture, and engineering.
A top-down racing game where an AI learns to drive using 7 distance sensors and nothing else. The car ended up so fast that it's physically impossible for me to stay on the track manually.
Read MoreThe real story of building an AI content generation pipeline for 27,000 pages — from writer agents and validator chains to a RAG system that cut costs by 87% and drove a +17% lift in impressions.
Read MoreA senior project in deep learning that started as a standard handwritten digit classifier — until I started throwing noise and rotation at it and learned that some problems are inherent to the data, not the model.
Read MoreA one-night personal project that turned into a lesson in prompt engineering. Take a photo, tile it into a grid, and have AI regenerate each section in a different artistic style — while somehow keeping the overall composition intact.
Read MoreA practical guide to structuring multi-agent systems with persona agents, validator chains, human feedback loops, and generation agents — lessons from building production agent pipelines.
