Summary
Jean-michel Sarr is a Research Engineer at Google with a PhD in Computer Science and a decade of experience building multilingual large-language-model datasets and evaluation benchmarks. He led curation of synthetic supervised fine-tuning data for Gemini 2.5 Flash, improving response quality across more than 25 languages, and has designed multilingual instruction-tuning and creativity benchmarks. His work blends rigorous academic research—publishing on complex data labeling and deploying ML models in environmental science—with product-focused engineering at scale. Based in Accra, he studies the real-world frictions PMs face, exploring how context-switching and repetitive processes erode strategic impact on small product teams. Comfortable moving between research, tooling, and teaching, he also developed core probability and statistics curricula and open teaching materials for Master’s programs. That Rare mix of multilingual ML systems expertise and hands-on domain modeling in resource-constrained settings makes him effective at turning messy, multilingual data into robust model improvements.
10 years of coding experience
2 years of employment as a software developer
Phd, Computer science, Phd, Computer science at Sorbonne Université
Bachelor, Mathematics, Bachelor, Mathematics at Université Paul Sabatier (Toulouse III)
Phd, Computer science, Phd, Computer science at Université Cheikh Anta Diop de Dakar (UCAD)
French, English