Engineering notes
Short write-ups from the work.
Backend, infrastructure and security. What we ran into, what we changed, and what we would do again. No bloat.
-
01 / 03
SQLite in production: WAL, backups and where it stops
The settings worth changing, the failure modes worth knowing, and the point where a single file stops being enough.
-
02 / 03
Idempotency keys for APIs that move money
How to make retries safe when a request can fail halfway, with a table layout and a request flow you can copy.
-
03 / 03
Zero-downtime deploys with systemd and a health check
Two instances, one health endpoint and a short script. Enough for a single host, no orchestrator required.