HS Lab

Technical experiments are reproducible run logs, not articles.

For embedded systems, OpenResty edge tuning, and Agentic AI, keep the commands, constraints, conclusions, and next steps instead of filling the page with invented metrics.

edgeOpenRestyOracle free tier

Doubling Oracle free-tier throughput by tuning OpenResty

An edge experiment on a tiny machine, broken down through worker binding, keepalive, Lua shared dicts, cache hit rate, and load-testing method.

embeddedfirmwaredebug

Reconstructing the boot path from UART logs

Build the smallest useful observation surface for firmware bring-up: reset reasons, boot stages, exception paths, and repeatable checkpoints.

agentsworkflowmemory

Treating agents as engineering pipelines, not chat boxes

Notes on how tool boundaries, context compaction, long-term memory, and checkpoints combine into a steadier personal R&D assistant.

embeddedlatencyDMA

Observing ring-buffer latency

Compare logs, GPIO toggles, and sampling windows to avoid optimizing the wrong problem with the wrong measurement method in embedded systems.

copied