Albin Soutif is an AI Engineer based in Madrid with seven years of experience at the intersection of machine learning research and engineering. He completed a PhD focused on continual learning—specifically investigating the causes of forgetting in neural networks—and now applies that expertise at Indra to build robust, privacy- and sustainability-minded ML systems. His background includes internships in adversarial ML at IBM and contributions to the popular Avalanche continual-learning library, where he improved and tested the ReplayDataLoader and clarified core loss criteria. Comfortable moving between theory and implementation, he has a strong mathematical foundation from top French and Spanish institutions. Colleagues value him for pragmatic refactoring that simplifies complex research code into production-ready components. He brings a rare mix of deep research insight and hands-on engineering to problems in online, unsupervised, and adversarially robust learning.
7 years of coding experience
5 years of employment as a software developer
Docteur en informatique, Artificial Intelligence, Docteur en informatique, Artificial Intelligence at Universitat Autònoma de Barcelona
Diplôme d'ingénieur, Mathématiques et informatique, Diplôme d'ingénieur, Mathématiques et informatique at Ecole Nationale Supérieure d'Informatique et de Mathématiques Appliquées de Grenoble
Classe préparatoire aux grandes écoles, Physique, Technique, Mathématiques, Classe préparatoire aux grandes écoles, Physique, Technique, Mathématiques at LGT JOLIOT CURIE (RENNES)
Avalanche: an End-to-End Library for Continual Learning based on PyTorch.
Role in this project:
ML Engineer
Contributions:21 reviews, 6 commits, 28 PRs in 1 month
Contributions summary:Albin primarily contributed to testing and refining the `ReplayDataLoader` within the Avalanche framework, a library for continual learning. Their work involved adding tests, including those to cover oversampling strategies. Additionally, the user made improvements to the `ReplayDataLoader` and incorporated options and documentation. Finally, there was a refactoring effort aimed at simplifying a core loss criterion.
Find and Hire Top DevelopersWe’ve analyzed the programming source code of over 60 million software developers on GitHub and scored them by 50,000 skills. Sign-up on Prog,AI to search for software developers.