Lean-stack engineering is an operating philosophy for small teams: one statically linked binary on one rented box, SQLite with write-ahead logging, few dependencies, and a bill small enough to read as an observability signal. Auditability is the deliverable; distributed-systems complexity is deferred until measurement proves it necessary.

Lean-stack engineering (Hanov, 2024; Ryer, 2024)

An operating philosophy for small teams that treats dependencies, runtime complexity, and infrastructure sprawl as recurring costs, and treats auditability plus operational transparency as the deliverable: one statically linked binary, one VPS, SQLite with WAL, and transparent billing as the first observability signal.

Hanov supplies the perimeter discipline; Ryer supplies the internal Go HTTP service shape (NewServer with explicit deps, handlers as closure factories, context cancellation at every layer) that keeps the binary inspectable as it grows.

For the longer treatment in narrative context, see Reliability: lean-stack engineering.

Part of Reliability on the learn hub.

See also

References

Steve Hanov, on running ten-thousand-dollar MRR businesses on a twenty-dollar monthly tech stack. The headline essay. Argues for one VPS, one statically linked binary, SQLite with WAL, and transparent billing as the operator's first observability signal.
Mat Ryer, on writing HTTP services in Go after thirteen years. The complementary axis: NewServer(deps...) with explicit arguments, handlers as closure factories rather than struct methods, context cancellation at every layer, and end-to-end tests of run() via httptest.

The techniques in these pages run in production behind spotinference's OpenAI-compatible endpoint. Get a key and try it: swap the base URL and the key in an existing SDK, and the first request streams back tokens.