Valentin Lacombe

A Research Engineer with a dual degree in applied mathematics and statistical learning, specializing in Natural Language Processing, Reinforcement Learning and Agentic AI. Building on a strong foundation in computer science and causal inference, recent work at INRIA involves advancing AI research and developing open-source tools for procedural data generation. Combines a rigorous analytical background with academic mentoring experience, driven to contribute challenging AI projects.

Research Engineer at INRIA laboratory (2025-01 – Present)

• Open-Source Development: Co-author and primary developer of reasoning_core, an open-source library that generates synthetic procedural data for LLM training purposes (SFT + RLVR) with project available on github and preprint available.
• Causal Reasoning: Implemented structural causal models within reasoning_core, observational, interventional, and counterfactual queries based on Judea Pearl's "Ladder of Causation."
• AI Automation & Innovation: Developed an AI-driven automation proof-of-concept (POC) in connection with the INRIA Startup Studio, leveraging n8n and LangChain to build and orchestrate intelligent, LLM-powered workflows.

High school teacher at Lycée Louis PASTEUR (2023-01 – 2024-12)

Full time replacement teacher in my former high school, Teaching mathematics and physics

Server at Pizza Bella (2022-01 – 2022-12)

Bank reception manager at Crédit Agricole (2021-01 – 2021-12)

Operator in a cheese factory at Eurial (2019-01 – 2019-12)

Orchard worker at les vergers de Muant (2018-01 – 2018-12)

Master of Science in Statistical Learning – Politecnico di Milano (2020-01 – 2023-12)

French graduate school of engineering in Major in Applied Mathematics – Ecole nationale supérieur des Mines de Nancy (2018-01 – 2023-12)