Backend and APIs

Data model, REST and GraphQL, queues, load, migrations. For teams that already have a frontend or a mobile app, and for systems that have hit the limits of their database.

NODE · GO · RUST

A backend written in a hurry works right up to the first real load. Database queries take seconds, background jobs block responses, migrations are run by hand on a Sunday night. Every new feature breaks an old one because the data model cannot carry it.

I start with the data model: entities, relations, what changes often and what is only read. Then the API — REST or GraphQL, with documentation the frontend can be written against in parallel. Heavy operations go to queues, hot data to a cache, search to an index rather than a scan. The core stack is NestJS and PostgreSQL; for compute and services under load, Go or Rust. Migrations live in code and run in CI before the release.

You get an API with documentation, tests and observability: logs, metrics, alerts. It can be handed to your team or connected to any frontend. FOQIN is built this way: a nightly exchange with 1C across 12,000+ items, part number and cross-reference search on an index. At Meetcap the processing queue holds the load when calls end at the same time.

What is included