Fabrice Lamarche He/Him

I am a PhD Candidate at Mila and Polytechnique Montréal, supervised by Prof. Amal Zouaq in the LAMA-WeST Lab. My research focuses on large language models applied to the medical domain, with a particular interest in information extraction, non-parametric memories and agentic workflows. In the Fall 2025 semester, I was also a teaching assistant for the NLP course at Polytechnique Montréal.

Before my PhD, I completed an M.Sc. in Computer Science (NLP) at the RALI, Université de Montréal, under the supervision of Prof. Philippe Langlais, on evaluation methods for open information extraction. That work led to BenchIE^FL, an information extraction benchmark published at ACL 2024, and to general work on evaluation methodology for information extraction and text generation systems, which remains a common thread across my research, from structured information extraction to hallucination detection.

Beyond academia, I have also held various industry positions, namely as a Scientist in Residence at IVADO, where I set up LLM fine-tuning and RAG for an EdTech startup, and as a speaker in IVADO’s AI awareness program, with talks such as The ABCs of AI and AI, a risk for our democracy?. I also worked as a Data Scientist at Intact Insurance and at Santé Québec.

News

  • 20 Oct 2026 — MedHal at AACL-IJCNLP 2026 Our dataset paper, MedHal, is accepted at AACL-IJCNLP 2026! It is now available in the ACL Anthology.

  • 12 Oct 2026 — UMCR on arXiv New preprint out on arXiv: UMCR, a pipeline for universal medical concept recognition with guideline integration and agentic linking. Check it out!

  • 01 Sep 2024 — Starting my PhD I started my PhD at Mila and Polytechnique Montréal, in the LAMA-WeST Lab, supervised by Prof. Amal Zouaq!

Selected publications

UMCR: Universal Medical Concept Recognition through Guideline Integration and Agentic Linking. Fabrice Lamarche, Amal Zouaq. arXiv preprint (arXiv:2610.17363). 2026. arXiv

MedHal: A Synthetic Dataset for Medical Hallucination Detection. Fabrice Lamarche *, Gaya Mehenni *, Neshat Elhami Fard, Odette Rios-Ibacache, Li Ming Wang, John Kildea, Amal Zouaq. Proceedings of the 15th International Joint Conference on Natural Language Processing and the 5th Conference of the Asia-Pacific Chapter of the Association for Computational Linguistics (AACL-IJCNLP 2026), Volume 1: Long Papers. 2026. * Equal contribution. ACL Anthology arXiv

Contact

fabrice.lamarche at mila.quebec

You can also find me on Google Scholar, GitHub and LinkedIn.