Vincent Cohen-Addad

Research Scientist at Google Research, NYC

I am a Research Scientist at Google Research based in New-York. My work focuses on the design of inference methods and reinforcement learning algorithm for training foundation models at better reasoning; and algorithms for optimization problems stemming from data analysis and machine learning.

I am very honored to be a joint recipient of the 2026 EATCS Presburger Award.

Google Scholar LinkedIn Download CV

Research Interests

AI-Assisted Math and Theoretical Computer Science

Developing novel reasoning and problem-solving harnesses for large-scale AI models to tackle advanced fundamental tasks. See for example our work for the International Math Olympiad 2025, or with the Paper Assistant Tool (PAT) at STOC'26, ICML'26, or NeurIPS'26 (see also this). See also our recent work on AI-based mathematical results

Reasoning

Improving reasoning of large language models through long-horizon data generation, and tailored training algorithms.

Algorithms for Optimization Problems

Designing efficient algorithms for optimization problems, with an emphasis on those arising in data analysis and machine learning.

Retrieval Augmented Generation

Enhancing large language models by retrieving and incorporating external knowledge to generate more accurate and factual responses (e.g., for AI Overview or AI Mode).