Now

What I'm currently up to. Last updated July 2026.

Learning

• Working through a book on LLM inference and training at scale — how big models are trained and served efficiently: parallelism, memory, arithmetic intensity, KV caches.
• Rebuilding my math foundations, aiming toward the math behind machine learning.
• Digging into how language models actually work under the hood, from attention to sampling.

Doing

• TODO