Melanie Fernandez Pradier

About

I am a senior researcher in Health Intelligence (opens in new tab) at Microsoft Research Cambridge since November 2020. My objective is to bring state-of-the-art machine learning (ML) approaches to the live sciences and healthcare. Currently, I develop deep generative models to help decode the immune system and diagnose diseases within the Antigen Map Project (opens in new tab), as part of Microsoft Immunomics (opens in new tab). My research interests include Bayesian modelling, interpretable ML, uncertainty estimation, and robustness under distributional shifts.

Previously, I was a postdoc at Data to Actionable Knowledge Lab (opens in new tab) – Harvard University advised by Finale Doshi-Velez (opens in new tab), working on interpretable ML, probabilistic modelling for impactful clinical applications. I primarily focused on developing human-centric ML models to personalize antidepressant prescriptions, and designing meaningful priors for deep Bayesian models.

As my background goes, I am originally a Telecommunication Engineer. I jumped into ML in industry, working on recommendation systems for two years at Sony Corporation R&D, first in Germany and then Japan. Afterwards, I got my PhD “Bayesian non-parametric models for data exploration (opens in new tab)” in Spain, with applications to sport sciences, economics and cancer research. On the personal side, I love Japanese culture, sports of any kind (e.g. ice-skating, yoga, dancing), and merge with nature in long-trail hikes.