When something breaks in production, the fastest way to recover is to understand why the system was built the way it was. We keep short decision records for every non-obvious choice — from database indexes to retry policies.
For small-to-medium services we still prefer Docker Compose + Nginx over Kubernetes. Less moving parts, easier debugging, and the same level of isolation for most workloads.
Bots and Mini Apps have become a full distribution channel. We treat them as first-class clients: proper auth, rate limits, background jobs and monitoring — the same standards we apply to web APIs.