I will ai systems engineer claude code, mcp, production deployment

F
formulaprofessi
F
formulaprofessi
Giulio O.
ai systems engineer claude code, mcp, production deployment
Full Screen
ai systems engineer claude code, mcp, production deployment

About this gig

Your prototype works on your machine. That is not the same as working on Monday morning, with real users, unattended.


Most AI-built code passes the demo and fails the first week: no timeouts, no retries, secrets in source, no tests, no cap on model spend.


I find those failures, rank them by what they will cost you, and fix them.


AUDIT - findings ranked by severity, each with file and line, why it matters and how to fix it. Plus a roadmap priced in hours, written so any engineer could execute it.


PRODUCTION HARDENING - everything above, then I do the work: timeouts, retries with backoff, secrets out of source and out of git history, tests on the critical path, CI on every push. Delivered as a pull request you review and merge.


PRODUCTION + DEPLOY - plus containerized deploy, structured logging, output caps on model calls, and handover documentation another engineer can act on.


I work on a fork or a branch, never your main branch. I never ask for production credentials and I do not need them. Mutual NDA offered before you ask, at no cost.


Send me the repository and I will tell you which tier it needs - or that it does not need one.



  • Application Type

    • Web Application

  • AI Type

    • Other

  • Programming Language

    • JavaScript
    • Python
    • TypeScript

  • Web Frameworks

    • React
    • Flask
    • Next.js

  • Desktop Frameworks

    • Electron

  • No & Low-Code Builders

    • Other

Get to know Giulio O.

Giulio O.

Production Engineer for AI Systems Audits, Hardening, MCP Servers

  • FromItaly
  • Member sinceFeb 2022

  • Languages

    Italian, English
I make AI systems survive production. Most AI-built code works on the happy path and fails everywhere else: no timeouts, no retries, secrets in source, no tests, no cost cap. It passes a demo and breaks the first week real users touch it. I find those failures, rank them by what they will actually cost, and fix them. Audits, hardening, MCP servers, autonomous agent loops. Python, TypeScript, Shopify. Engagements run from a fixed-scope audit to a multi-week build. Every one ships with tests, documentation and a security handover. Read-only on audits. I never ask for production credentials.

My Portfolio

Related tags