Senior AI Engineer

This is a fully remote position, open to applicants in France.

📋 Description

• Develop and fortify agentic loops in production, encompassing LLM orchestration, tool invocation, streaming, event handling, error recovery, checkpointing, and resumability.

• Tackle security, permissions, and costs with agentic RBAC, audit trails, isolated execution, and token and latency budgets.

• Oversee context and memory engineering, which includes context creation, retrieval, compaction, and both short- and long-term memory management.

• Transform business capabilities into clearly defined, documented, versioned, and tested tools made available internally and through MCP.

• Direct agent behavior via prompts, instructions, skills, planning, tool selection strategies, guardrails, and anti-prompt-injection measures.

• Set up evaluation practices utilizing golden datasets, LLM-as-judge, human assessments, regression tracking, and error analysis.

• Collaborate with Product teams and domain experts in accounting on projects like ComptAssistant, Studio Assistant, MCP, document extraction, Autopilot bookkeeping, and revisions.

• Keep track of AI trends, cutting-edge models, emerging architectures, multimodal AI, and optimization methods.

• Manage roadmap items and deliver agentic use cases from start to finish in production, complete with evaluation sets and quality metrics.

• Enhance the agent harness, evaluation practices, and AI platform in collaboration with engineers and data practitioners.

• Share insights and best practices, and potentially recruit and mentor team members.

⛳️ Requirements

• 5–8 years of relevant experience.

• Exceptional skills in Python.

• Practical experience in building LLM and agentic systems at scale within a production environment.

• Proficiency in prompting, tool usage, context creation, RAG, failure management, state handling, and reliability.

• Familiarity with golden datasets, LLM-as-judge, human evaluations, A/B testing, regression, and edge-case assessments.

• Strong understanding of applied LLM/ML and AI infrastructure, including model serving, vector databases, and considerations of cost and latency.

• Proficient in English.

• Ability to communicate effectively with non-technical domain experts.

• Nice to have: experience with model fine-tuning and post-training techniques such as SFT, DPO, and RL.

• Nice to have: familiarity with MCP.

• Nice to have: familiarity with Ruby.

• Full work authorization in France or an eligible European country.

• No visa sponsorship is available.

🏝️ Benefits

• 25 vacation days covered by Pennylane.

• Competitive compensation package.

• Company share options.

• Budget for home workspace setup.

• Monthly allowance for coworking spaces.

• Access to 8,000 fitness locations across Europe and over 300 wellness activities through Gymlib.

• Busuu access for enhancing English or French language skills.

• Latest Apple equipment provided.

• Remote work allowed from an eligible European country within a maximum two-hour time difference from CET.

• Company events, including Tech Days every two months and an annual seminar.

• For employees based in France: 6 to 12 RTT.

• For employees based in France: 5 weeks of PTO.

• For employees based in France: lunch credits via Swile.

• For employees based in France: healthcare coverage through Alan Blue.

• For employees based in France: regular events in cities like Lyon, Bordeaux, Nantes, and others.