Skip to content

Services

I'm Artur Shiriev, the maintainer of modern-python. Besides the open-source work, I help Python backend teams that use coding agents keep production stable: I make the codebase easier for agents to work in, review where delivery stands, and lock down what agents can reach.

The libraries stay free and MIT-licensed whether or not you hire me. You don't have to use them either. When one fits, it saves time; when it doesn't, I work with what your team already runs.

Agent-ready backend

For teams where coding agents write a growing share of the code. A fixed-scope sprint that ends in merged pull requests in your repositories:

  • Context the agent can use. AGENTS.md, skills, and architecture rules written where the agent reads them and checked in CI, with a LikeC4 model of the system when the boundaries need one.
  • Checks for agent-written code. Lint, type, test and coverage gates plus structural tests and hooks, run the same way locally and in CI, along the lines of the org's own standard.
  • Access to production signals. CLI and MCP access to your error tracker, logs, metrics and traces, so an agent can check a change against how the service actually behaves.
  • Telemetry worth querying. Structured logs, traces, metrics and error tracking in FastAPI, Litestar or FastStream services, usually with lite-bootstrap, so people and agents can debug from the same data.

The modern-python repos run on the same setup, minus the production parts.

Delivery health check

A short, fixed-scope review, and a good first step before a sprint. Automated scanners for dependencies, vulnerabilities and code smells run first, and I treat what they find as raw input. Most of the time goes to what the tools can't see:

  • recent outages and incidents, what caused them, and which ones keep coming back
  • whether your logs, traces and metrics are good enough to debug from, for people and for agents
  • whether code review keeps up with how much code the agents produce, and how deployment frequency, change failure rate and time to restore have moved since the team adopted them
  • CI, the release process, and how a change reaches production
  • whether the architecture docs still match the code

You get a written report with findings ranked by impact and effort, a plan for the first fixes, and a call to go through it with your team. The review doesn't change your code. When it calls for design changes, such as dependency injection with modern-di, a transactional outbox with faststream-outbox, or resilient HTTP clients with httpware, I can make them as follow-up work.

Agent security and governance

For teams rolling coding agents out beyond a few early users, or unsure what the agents can touch:

  • a usage policy: which tools and repositories, which changes need human review, and the size and test requirements for agent pull requests
  • permissions: which commands agents may run, which secrets and tokens are within their reach, and whether anything reaches production
  • MCP servers and other integrations: which to trust and what data they see
  • prompt injection through issues, pull requests, docs and CI, wherever agents read text they didn't write
  • a rollout plan: one team first, then wider once the guardrails hold

Library support

If you run modern-python libraries in production, I can help with migrations (for example from that-depends or dependency-injector to modern-di) and with incidents that involve one of them.

Get in touch

Email lesnik512@gmail.com with a few lines about your team, your stack, and what you'd like fixed. I'll reply with questions or a proposal for scope and timing.