Senior AI Engineer 🇪🇺

Role overview Join one of Europe's fastest-growing fintechs as a Senior AI Engineer building the agentic layer of a financial operating system used by small businesses and accounting firms across France and beyond. The role focuses on productionizing LLM-powered agents, the harness around them, and the evaluation that proves they work on real accounting workflows. You will join one of three ML and AI teams (Capabilities and Infrastructure, Agent Behaviour, or the Studio Assistant), each owning a distinct slice of the AI platform.

Responsibilities - Build and harden agentic loops in production, including LLM call orchestration, tool calling, streaming, error recovery, checkpointing, and resumability, with security, permissions, cost, and latency in mind. - Own context and memory engineering, including context construction, retrieval, compaction, short- and long-term memory, and the latency-versus-cost trade-offs they imply. - Turn business capabilities into well-specified, documented, versioned, and tested tools, exposed internally and via MCP, and steer agent behaviour through system prompts and instructions. - Contribute to shared foundations such as agent harness, agentic runtime and sandboxing, model gateway, tool registry, guardrails, observability, and LLM FinOps. - Develop evaluation methodology and harnesses, and later explore domain tuning and model fine-tuning for safe, measurable agent behaviour. - For the orchestrator agent, work on intent understanding, tool selection, job launch, report production, and asking for human validation when appropriate.

Requirements - Strong software engineering background combined with applied LLM and agentic systems experience at a senior level. - Production experience building agentic systems, including orchestration, tool calling, error recovery, and checkpointing. - Solid grasp of context and memory engineering, retrieval, and the trade-offs between latency, cost, and quality. - Care for security and permissions in AI systems, including agentic RBAC, audit trails, isolated execution, and token or latency budgets. - Comfort designing evaluation harnesses and measuring real workflow impact, not just offline benchmarks. - Ability to deliver at scale with high trust, given the financial sensitivity of the workflows involved.

Benefits and work setup - Remote-friendly culture with significant distributed teams across Europe and an international environment of more than 25 nationalities. - Strong emphasis on trust, proactivity, and autonomy, with an engineering organization of roughly 1,000 people and growing.