aykol / tensorflow

Computation using data flow graphs for scalable machine learning

Hello! I am currently a Research Scientist at Google DeepMind.

Previously, I led a Data Science team at Rivian, focusing on ML & analytics for batteries. Prior to Rivian, I spent a little over 4 years at Toyota Research Institute (TRI) as a researcher.

Most of my recent published work is centered on combining machine learning with physics and chemistry to accelerate R&D. Among a few, these two deserve the spotlight as open-source Python libraries:

🚀 github.com/TRI-AMDD/CAMD: an end-to-end autonomous computational platform for closed-loop optimization. It was the Bayesian optimization & workflow engine behind a few papers: 1, 2, 3.

🚀 github.com/TRI-AMDD/piro: a recommendation system that combines physics (of nucleation) with ML-inspired approximations to find feasible synthesis routes for compounds. Check out this paper to learn more.

🤔 My research & intellectual interests these days cover Bayesian & closed-loop optimization methods, physics-informed ML algorithms, and on the materials side predictive synthesis and discovery. I'm also fascinated by network science as a field.

⚡ Fun fact: Looks like I get to update this github account every few years!