Current position: Lead data scientist at Fuelbetter Technologies (London, UK).
I obtained my PhD in Applied Mathematics in 2013, working on splitting methods for Langevin Dynamics (see my thesis here). I was a postdoc in Statistics at The University of Chicago (2014-2018) and a postdoc in Machine Learning methods at The University of Edinburgh (2018-2019).
I enjoy working on practical problems at the intersection of mathematics and computer science. In the past I’ve worked on novel integration methods for molecular dynamics, Bayesian sampling schemes for cosmology, modeling cascade failures in power networks, and training neural networks for renewable energy.
|Feb 28, 2021||The Fuelbetter app is live on Apple iOS! Nice to have the data I’ve been working on available for public consumption (pun intended). It still amazes me how accurate the machine learning reverse-engineering can be.|
|Oct 11, 2020||Working on a new side-project RACECAR, used for learning a noisy gradient correction on-the-fly.|
|Apr 4, 2020||Officially started working at Fuelbetter Technologies as the lead in machine learning and data science, based in London. The role involves developing algorithms for solving statistical inverse problems at scale, by writing and deploying code that reverse-engineers food nutrition for over a million packaged food items.|
- Molecular Dynamics with Deterministic and Stochastic Numerical Methods2015
- Langevin Markov Chain Monte Carlo with stochastic gradients2019
- Umbrella sampling; A powerful method to sample tails of distributionsMonthly Notices of the Royal Astronomical Society 2018