Paris, Île-de-France, France
ML scientist at Owkin. Focused on developing AI-powered solutions that drive breakthroughs in human health. Past experiences and current work include: • Representation learning on single-cell & spatial-omics (contributed to 3 papers and 3 posters) • Generative & diffusion models for proteins - diffusion-style models for binding-site / cryptic-site identification and pocket–binder matching, leveraging 3D structural information. • Target validation and druggability optimisation (poster at ESMO) • Clinical-trial optimisation - 2 multimillion-$ BMS projects (contributed to 2 papers) • Dockerized PyTorch pipelines on AWS, delivering models into pharma workflows (incl. BMS trial-optimisation projects) • Time-series RNNs predicting ICU cardiac arrest • Advised national tech policy connected with national electronic health records • Raised $100 k pre-seed and led 5-person cross-functional teams I enjoy owning problems end-to-end: from open-ended research questions, to experimental design, to robust, production-ready code that real teams can use.
Develop deep learning models for protein sequence–structure–function, including GNNs/Transformers for protein-pocket matching, cryptic-site detection and druggability prediction. Design and test generative / diffusion-style models for binding-site matching and cryptic-site identification, using 3D structural information to guide design hypotheses. Multimodal rep-learning (single-cell, spatial-omics); clinical-trial optimization. Contributed to 3 publications + 3 posters (incl. ESMO). Built pocket-centric GNN/Transformer pipelines used in target validation. Supported BMS programs with trial-optimization models (2 papers contributed).
The Kairos Society pushes the world forward. It is a global community of top global leaders using entrepreneurship and innovation to solve the world's greatest challenges. Supporters include Bill Clinton, Richard Branson and Peter Diamandis.
Visiting Researcher in Machine Learning for Healthcare — focused on ICU data analysis in collaboration with Boston Children’s Hospital. Worked on developing and evaluating machine learning models for critical care and patient outcome prediction. Awarded research fellowship. Supervisors: Prof. Mauricio Santillana & Prof. Leo Celi.
Research in Machine Learning for Healthcare — continued from Master’s thesis. Worked on a recurrent neural network (RNN)–based model for real-time blood pressure prediction and anomaly detection in the ICU setting, aimed at clinical implementation. Supervisors: Prof. Sebastian Goodfellow & Prof. Mjaye Mazwi.
Heritage is on a mission to transform investing for families, entrepreneurs, and leading advisors, through modern tech, superior portfolio construction, and access to top alternative assets. Worked on the early stages helping design and set up from scratch the tech vision and tech stack