Ingénieur(e) IA / DevOps – Industrialisation d’agents IA et de serveurs MCP (IT) / Freelance

Job summary

Industrialize an existing AI-agent prototype for production, improving the quality, scalability, maintainability, and operational readiness of an orchestration agent and three MCP (Model Context Protocol) servers without adding business features or changing the approved technical architecture. The role includes Python refactoring, automated testing and CI/CD, performance and resilience improvements, observability and operational security, plus documentation and knowledge transfer. Work is carried out within an agile squad using technologies such as pytest, Docker, Kubernetes, GitLab CI, Locust or k6, and Claude Code.

Qualifications

• Three to seven years of experience scaling software components.
• Software engineering experience in AI focused on code quality, robustness, and operability.
• Advanced Python development in production environments.
• Experience with software industrialization, automated testing, and CI/CD.
• Experience with performance and load testing, observability, and DevOps practices.
• Knowledge of AI agents, MCP, and LLM orchestration frameworks.
• Able to work rigorously and autonomously with existing code and defend technical decisions in reviews.
• Professional technical English for documentation and written communication.
• Banking or regulated-environment experience is an advantage.

Responsibilities

• Diagnose the four components and prepare a prioritized, estimated upgrade plan within five days.
• Refactor and structure the code, externalize configuration, and improve error and retry handling.
• Implement unit, integration, and contract tests; integrate them into CI/CD and define a coverage threshold.
• Improve scalability using concurrency, asynchronous calls, connection pooling, timeouts, retries, and circuit breakers.
• Run performance and load tests and document target sizing and known limitations.
• Implement structured logs, metrics, distributed traces, dashboards, and alerts, including token and inference-cost tracking.
• Strengthen secrets management, authorization controls, and audit logging.
• Produce operational documentation, diagnostic procedures, and a runbook, then transfer knowledge to the internal team.
• Use a coding assistant judiciously and critically review generated code to prevent regressions.

Skills

Languages

Work schedule

Industry

Contract length

Relocation