Technical articles on production LLM systems: data integration, reliable agents, evaluation, and recovery.
A practical design for AI workflow engines with durable state, deterministic recovery, retries, audit trails, evals, and cost control.
Where the calculator analogy for LLMs holds and where it breaks: what interpretability, chain-of-thought and philosophy of mind say about thinking.
An engineering report on porting a small language model to an Xbox Series S: what runs at 71 tok/s on the Zen 2 CPU, updated with the measured GPU verdict.
Five disciplines from banking security — durable state, deterministic recovery, dual control, and audit trails — applied to LLM agent evaluation.