AI Agents SDLC Developer

Project overview

As a developer within the Data & AI team, the AI Agents SDLC Developer builds, deploys, and maintains AI agents and multi-agent orchestrations for engineering programs.

Starting from an existing agent framework and an agent library, the developer extends and improves these assets to make them robust and reusable tools, adapted to the engineering processes and practices of each program. The position combines hands-on software development with close collaboration with V&V engineers, program managers, and developers, within the approved environment including GitHub Copilot, GitLab Duo, and associated AI tools.

Key Responsibilities

Mandatory Skills

Desired Skills

Work Environment

About emagine For 30 years, our clients have seen emagine as a reliable partner to whom they entrust their IT, Application, Digital, and Cybersecurity projects. Present in France through 5 offices and in ten other European countries, we now form a group with a turnover of €600 million, with over 4,500 consultants on assignments. We support our clients with high value-added services and offer complete audit and external resource management solutions. We owe our ability to undertake complex or specialized assignments to our spirit of innovation and the agility that emagine places at the heart of its strategy.

• Build agents: design, develop, test, and deploy AI agents supporting engineering activities throughout the SDLC, including requirements analysis, development, code review, verification and validation (V&V), and documentation.
• Build orchestrations: design multi-agent workflows, chain agents, route tasks, manage shared context and handoffs, handle failures, and retain human review and approval steps when necessary.
• Extend existing assets: understand the existing agent framework and agent library, reuse their interfaces and models, and improve capabilities and reliability rather than starting from scratch.
• Work with siloed teams and programs: consider different processes, priorities, and maturity levels, identify common needs, and consolidate reusable capabilities without losing program-specific requirements.
• Collaborate with V&V engineers, program managers, and developers: facilitate workshops, understand engineering tasks and constraints, and translate them into agent behaviors, integration requirements, and testable acceptance criteria.
• Integrate engineering tools: develop within GitHub Copilot, GitLab Duo, and other approved tools environments; connect repositories, CI/CD pipelines, and IBM ALM RDNG, EWM, and ETM via supported APIs, MCP (Model Context Protocol), and A2A (Agent-to-Agent) protocols, where applicable.
• Industrialize and maintain agents: apply automated testing, reproducible evaluations, version control, packaging, deployment, and monitoring; apply the principle of least privilege access, protect confidential data, maintain traceability between requirements and tests, and document operations and knowledge transfer.
• Master's degree in Computer Science or Software Engineering.
• Proven hands-on experience in building and deploying AI agents and multi-agent orchestrations beyond proof-of-concepts, including extending an existing agent framework and agent library.
• Hands-on LLM-based development: agent frameworks, tool calling, Retrieval Augmented Generation (RAG), prompt engineering, and context engineering.
• Strong software engineering fundamentals: Python, API development and integration, Git, automated testing, packaging, and maintainable code.
• Hands-on experience with GitHub Copilot, GitLab, and GitLab Duo, CI/CD workflows, and secure agent integration; understanding of MCP and A2A protocols and the capabilities of the respective platforms.
• Experience in adapting SDLC agents to engineering programs and V&V practices, including program-specific processes and acceptance evidence items.
• Experience with GitHub Copilot, GitLab Duo, Azure AI Foundry, Azure OpenAI, or similar approved services for agent development and deployment.
• Hands-on experience implementing MCP servers or clients, A2A interoperability, and connectors to IBM ALM or other engineering tools.
• Experience in evaluating the reliability of agents and multi-agent systems, managing prompt injection risks, implementing human approval controls, and supporting Responsible AI compliance.
• Experience in operating reusable agent libraries, including configuration management, observability, version management, and adoption by multiple teams.

Project details

• Start: 15/11/2026
• Duration: 36 mois
• Remote: hybrid
• Location: 93000, Bobigny, Île-de-France, FR
• Rate: 73-78 EUR/h
• Languages: French