Grigori Fursin

My R&D playground    Mission    LinkedIn    Google scholar    Twitter    Medium    GitHub    EMail

Co-designing more efficient and cost-effective AI systems at FlexAI | supporting open science, reproducibility and automation at ACM, IEEE and MLCommons | ex VP of MLOps at OctoAI (now Nvidia) | ex co-director of the Intel Exascale Lab | ex senior tenured scientist at INRIA | ex adjunct professor at the University of Paris-Scalay | PhD from the Unviersity of Edinburgh


My name is Grigori Fursin, and I currently live in the suburbs of Paris. I was among the first researchers to pioneer the use of AI and ML to modernize computer systems—including compilers, run-time systems, software, and hardware—contributing to more efficient, cost-effective, and scalable solutions for AI, ML, and other emerging workloads while managing their growing complexity and reducing time to market. I have also been an active advocate for open science, reproducibility, and open-source contributions since 2008, when I released all my research code, data, models, and experiments for our ML-based self-optimizing compiler to foster collaborative and reproducible R&D to co-design more efficient AI and ML systems: ACM TechTalk'21.

After serving as a senior tenured research scientist at INRIA, an adjunct professor at the University of Paris-Saclay, and co-director of the Intel Exascale Lab, I transitioned my research and open-source tools into industry. I founded several successful companies in the fields of performance optimization and knowledge management, the most recent of which was acquired by OctoAI (now Nvidia), where I served as VP of MLOps.

As part of my community service, I helped establish artifact evaluation and reproducibility initiatives at ACM and IEEE conferences,and introduced a unified artifact appendix adopted by ASPLOS, CGO, PPoPP, SuperComputing, MICRO, and other conferences. I also contributed to setting up the Intel Exascale Lab, the non-profit cTuning Foundation, the educational Collective Knowledge initiative and MLCommons to accelerate AI innovation for the benefit of all. I was honored to receive the ACM CGO Test of Time Award, multiple Best Paper Awards, the INRIA Award for Scientific Excellence, and the EU HiPEAC Technology Transfer Award for my research. I'm also very glad that my open-source technology helps many companies and organizations, including MLCommons.

With an interdisciplinary background in computer systems, compilers, machine learning, physics, and electronics—along with over 20 years of experience in research, development, and industry—I help companies, startups, universities and non-profits establish R&D labs and launch innovative projects with solid methodologies for collaborative and reproducible R&D.

I also regularly share my knowledge, expertise, and wisdom with students, researchers, businesses, governments, and investors, helping them navigate the complex and rapidly evolving deep-tech landscape, avoid common pitfalls, and drive meaningful progress.

Please check a few recent presentations and publications if you want to learn more about my current and past activities: ACM TechTalk'21, Google scholar, keynote at ACM REP'23, ArXiv white paper'24, overview in Philosophical Transactions of the Royal Society'21, ArXiv paper'17 about ML-based compiler auto-tuning, and my reproducibility initiatives at ML and Systems conferences since 2014.

My current activities:

Brief summary of my past activities:

My timeline (may not be up-to-date):