AI Software Engineer (F/H/NB)

SFEIR is a technology expertise firm specializing in the design, architecture, and industrialization of custom digital solutions. We support organizations in their most complex transformations by combining the rigor of Software Craftsmanship with mastery of future technologies. Convinced that digital sovereignty is a major strategic issue, we design secure, high-performance, and independent architectures, supported by strong partnerships with key players like SENS and Scaleway. As a reference player in artificial intelligence, we are evolving modern engineering standards. Through our GenAI RAISE platform and our status as an Anthropic Preferred Partner (within the Claude Partner Network), we place Context Engineering and generative AI at the heart of our practices.

Our key figures:

• 950 consultants passionate about tech and augmented by AI.

• 8 agencies in France, Luxembourg, and Belgium.

• Anthropic Select Partner within the Claude Partner Network.

• 15 years of partnership with Google Cloud, AWS, Microsoft, Databricks, Confluent...

• +5000 people trained annually via the SFEIR Institute.

Joining SFEIR means integrating a collective where technical culture prevails. We offer dedicated career paths for experts (Individual Contributor / IC), allowing you to progress in high technical mastery without having to switch to organizational management. We integrate AI into our own engineering practices (daily use of RAISE and Claude Code) to automate low-value-added tasks and focus our efforts on designing resilient architectures.

▪️ About the role

You will design and deploy AI agents and complete agentic systems executing end-to-end business processes: advanced customer support, multi-source search, IT operations, as well as advanced use cases related to the industry of the future and digital continuity (e.g., requirements traceability, predictive maintenance, engineering continuity and digital twins). You know how to develop custom agents in code as well as configure agents via low-code/no-code platforms with platforms like Gemini Enterprise.

Your field of action combines technical mastery of the software development lifecycle (SDLC) and a fine functional understanding of business challenges. You will orchestrate agent autonomy, planning, tool selection, and memory, while ensuring reliability, governance, and adherence to guardrails.

The missions:

• Analyze and understand business needs and processes with a clear functional vision to define automation use cases and task allocation (agentic autonomy vs. human validation).

• Design and develop complete agentic systems (mono- or multi-agent), choosing the appropriate approach: custom development or programming/configuration of low-code/no-code agents (Gemini Enterprise, Vertex AI Agent Builder, AWS).

• Build agent tools: function calling, API and information system integration, creation and consumption of MCP servers, asynchronous and long-duration workflows.

• Master reliability: error and recovery management, control loops, budget and step limits, behavior in the face of the unexpected.

• Evaluate agents: scenario sets, trajectory analysis, task success rates, LLM-as-judge, regression tests with each change of prompt, tool, or model.

• Monitor in production: detailed logs, costs per task, drift and loop detection, dashboards.

• Secure autonomy: minimal permissions on tools, protection against prompt injection, isolation of sensitive actions, human validation, traceability, compliance (GDPR, AI Act).

• Integrate into the complete software development lifecycle (SDLC) and deploy on the cloud (Google Cloud / Vertex AI) with CI/CD integration and cost management.

• Share: document patterns, train client teams, contribute to SFEIR publications and events.

▪️ The ideal candidate

• 5 years of experience in software development (excluding internships and apprenticeships) with a cross-functional SDLC vision and solid initial experience on LLM / AI projects
• Demonstrated practice in creating AI agents: custom development (frameworks/code) or creation via low-code/no-code environments (Gemini Enterprise, AWS,...)
• Knowledge of MCP and inter-agent interoperability protocols (A2A)
• Fine understanding of agent failure modes: drift, loops, incorrect tool selection, loss of context
• Strong mastery of APIs, asynchronous programming, and event-driven architectures
• Engineering rigor: testing, observability, reviewed code
• Interpersonal skills and good functional level to interact with business teams, understand their challenges, and explain the possibilities and limitations of AI agents

A plus

• Experience in agent evaluation and observability (OpenTelemetry, Langfuse…)
• Knowledge of the BMAD methodology
• Google Cloud (Vertex AI, Cloud Run, Gemini),
• Security awareness: threat models for agents, identity and permissions management
• Notions of RAG, viewed as an agent tool
• Open source contributions, agentic side projects, talks