Field notes on engineering systems that actually scale.
643 articles written by the engineers, designers and product strategists building production systems at Sijenius. No fluff, no recycled takes — just lessons from real builds.
The deep-dive we recommend this week.
Designing event-driven systems that survive their first Black Friday.
A 9,400-word teardown of the messaging patterns, idempotency rules and back-pressure strategies we use when a client processes 40M events an hour and a single duplicate would cost them $180k. Includes four full architecture diagrams and the actual retry policies we ship.
Read the full piecePublished this month.
Cutting p99 latency by 71% on a Postgres-heavy workload
Connection pooling, partial indexes, and the surprisingly large impact of switching from JSONB to typed columns.
The eval harness we use before any model ships to production
Twelve automated checks, three human-in-the-loop review gates, and the metric we trust more than accuracy.
Why we killed our 400-component library and started over
The case for fewer primitives, looser opinions, and treating your design system as a product — not a deliverable.
A pragmatic guide to feature flags without the vendor lock-in
Open-source stacks, the rollout patterns that don't page you at 3am, and how to clean up dead flags automatically.
What "good" looks like in a product discovery sprint
The five interviews, the three artefacts, and the one meeting that decides whether to keep building or kill the bet.