What I am learning now
Working notes on AI agents, distributed systems, and learning through small experiments.
My current learning is split between two areas that increasingly overlap: AI agents and dependable distributed systems.
AI agents
I am interested in the gap between an impressive demonstration and a useful, repeatable workflow. That means paying attention to context design, tool boundaries, evaluation, and the ways a system recovers when an assumption is wrong.
Small experiments are the fastest way for me to build intuition. I prefer a narrow workflow with a clear success condition over a broad assistant that is difficult to evaluate.
Distributed systems
I continue to study the practical side of distributed infrastructure: backpressure, idempotency, observability, and capacity planning. These ideas are not new, but they become newly relevant whenever a system changes scale or ownership.
The most valuable lessons often come from understanding how a system behaves outside its happy path. Good architecture makes failure visible, bounded, and recoverable.
Learning in public
Writing is part of the learning loop. Turning an experiment into a short note forces me to identify what I actually understand and what still feels vague.
These notes will evolve as projects produce better questions. The aim is not to sound finished; it is to leave a useful trail of decisions, mistakes, and updated mental models.