Benjamin Muskalla

Writing

Blog

Notes on software, tooling, open source, and the odd puzzle — collected over the years.

Recent

  • Sep 26, 2026
    OpenAPI specs as RL gyms

    We turned OpenAPI specs into mock services for RL training and improved a tool-use benchmark by about 28%; most of the work went into turning a simple seed spec into a world complex enough to be worth learning from.

  • Sep 17, 2026
    The half-life of engineering best practices

    AI agents aren't just changing how fast we write software. They're changing the economics behind decades of engineering wisdom.

  • Sep 13, 2026
    GOTOken: An Inference Engine in QBasic

    Running SmolLM2-135M with an inference engine written entirely in QBasic.

From the archive