I believe in the use of challenges, competitions and benchmarks to find solutions to complex data science problems.
ML Challenges: Result Submission VS Code SubmissionMany competitions are organized each year on platforms such as Kaggle, CodaLab Competitions or EvalAI. Some allow you to upload your model’s predictions, while others allow you to upload directly your...
A brief history of competitions in machine learningThe idea of leveraging a community of experts and non-experts to solve a scientific problem has been around for hundreds of years. The long history of scientific challenges begin with...
Designing a Data Science Competition is an Excellent Way to LearnOver the past few years, we have explored the benefits of involving students both in organizing and in participating in challenges as a...
Beyond Accuracy: Exploring Exotic Metrics for Holistic Evaluation of Machine Learning ModelsMachine learning has undoubtedly become a powerful tool in today’s data-driven world, but are we truly tapping into its full potential? Traditional evaluation metrics like...
AutoDL is a series of Automated Machine Learning (AutoML) competitions applied to various domains such as image, video, text, speech, and tabular data, where Deep Learning excels.
AI4Industry Challenge, in collaboration with Dassault-Aviation and the Paris Region, is an ambitious time series regression contest from aeronautical data for predictive maintenance, where Gradient Boosted Trees methods have excelled.
L2RPN 2023 addresses climate change by promoting the energy transition to renewable sources like solar and wind. Organized with Réseau de Transport de l'Électricité and the Paris Region, participants manage a simulated power grid.
CodaLab Competitions is a versatile, open-source web platform hosting machine learning competitions, widely recognized in academia and industry.
Codabench is an advanced version of CodaLab Competitions, offering a faster and more responsive experience and rapidly becoming an influential platform for benchmarks and competitions.
"Methodology for Design and Analysis of Machine Learning Competitions", PhD Thesis, 2023.
"AI Competitions and Benchmarks: the Science behind the Contests", Book preprint submitted and under review at Data-driven Machine Learning Research (DMLR).
"Codalab Competitions: An open source platform to organize scientific challenges", Journal of Machine Learning Research (JMLR), 2023."Filtering participants improves generalization in competitions and benchmarks", European Symposium on Artificial Neural Networks (ESANN), 2022.Welcome to my personal website! I have gratuated from University Paris-Saclay in computer science and machine learning in 2018. Right after that, I became an independant freelancer and worked closely with INRIA, Chalearn and Google, to organize challenges in data science. Between 2020 and 2023, I served as a research engineer and PhD student at LISN, under the supervision of Isabelle Guyon, also known as the "Queen of Challenges" and co-inventor of SVM. My thesis discusses the methodology and experimental design in machine learning, the performance comparison between models and the organization of competitions. From this period, I started having a role in the development and the administration of CodaLab and Codabench, two major AI competition platforms. I also organized two competitions, each featuring a €500,000 prize, in partnership with French industries, Dassault-Aviation and RTE. In this context, I developed a diverse skill set, including system administration, software engineering, data analysis, scientific research, writing, management and communication. I tend to treat the field of artificial intelligence as an experimental science, and to get my hands dirty by directly diving into implementation. I am also a music lover, producer and composer. Thanks for reading!
