Posts
All the articles I've posted. 46 posts, page 2 of 12.
How do you know your agent works?
Published: at 02:40 AMScore by outcome, beat doing nothing, beat a random policy, ablate every part, and count the cost. A practical checklist for evaluating AI agents.
The model is half the product. The harness is the other half.
Published: at 09:30 PMWhy the code around a language model decides so much of how an agent behaves, and why its contribution is often hard to separate from the model's.
Experiment, build, understand: how I pick what to work on
Published: at 03:00 AMThe thread behind my research, hackathons, voice apps, quantum lessons, and painting, and what this blog is going to be about.
What I'm looking into this year
Published: at 02:30 AMFour questions I'm spending senior year on: world models, optimizers, how to evaluate AI harnesses, and whether a model can know when it's wrong.