Miguel Hernández-Lobato

Timeline

1. 2004–2010

   Computer science in Madrid

   He earns a B.Sc. in Computer Science at Universidad Autónoma de Madrid in 2004, with a special prize for the best academic record on graduation, and an M.Phil. in 2007. He completes his PhD there in 2010 with a dissertation on semi-parametric methods and sparse linear models, advised by Alberto Suárez González.

2. 2011–2016

   Postdoc at Cambridge and Harvard

   From 2011 to 2014, he is a postdoctoral researcher in Zoubin Ghahramani's Machine Learning Group at Cambridge. From 2014 to 2016, he is a postdoctoral fellow in Ryan Adams's Harvard Intelligent Probabilistic Systems group, funded by a Rafael del Pino Foundation fellowship.

3. 2014–2015

   Entropy search and probabilistic backpropagation

   In 2014, he coauthors Predictive Entropy Search, a Bayesian optimization method that picks the next point to evaluate by its expected information gain about the global maximum. In 2015, he and Ryan P. Adams present probabilistic backpropagation, a scalable way to learn Bayesian neural networks by propagating probabilities forward through the network and gradients backward.

4. 2016–2022

   Lecturer to professor at Cambridge

   Cambridge's Machine Learning Group announces in April 2016 that he will join as University Lecturer in Machine Learning, and he starts in the Department of Engineering in September 2016. He becomes Associate Professor in October 2021 and Professor of Machine Learning in October 2022.

5. 2018

   A continuous representation of molecules

   In 2018, he coauthors an ACS Central Science paper with Rafael Gómez-Bombarelli, Alán Aspuru-Guzik and others, in which a neural network turns molecules into continuous vectors. The continuous representation lets gradient-based optimization guide the search for better molecules.

6. 2020

   Turing fellowship and ELLIS unit

   He becomes director of the Cambridge ELLIS unit in September 2020. In November 2020, he is awarded a Turing AI Acceleration Fellowship for machine learning for molecular design, aimed at generative models that place atoms in 3D space and provide synthesis information.

7. 2023

   An ELLIS program on molecules

   ELLIS accepts Machine Learning for Molecule Discovery as a new ELLIS Program in 2023. He coordinates the proposal with Günter Klambauer of Johannes Kepler University Linz and Nadine Schneider of Novartis, and ELLIS lists him as its Program Director.

8. Since October 2025

   Full professor and industry roles

   He becomes Full Professor of Machine Learning in the Department of Engineering at Cambridge in October 2025 and continues as director of the Cambridge ELLIS unit. He is Head of Research at Boltzbit Limited since September 2025 and Chief AI Officer at Angstrom AI since November 2024.