About me

Unlike the archetypical chemistry student, I decided to spend most of my time at university working on theoretical and computational problems.

For my bachelor thesis, I joined the Computational Biochemistry Unit at the University of Oxford. Subsequently, I was fortunate to join Tristan Bereau’s group formerly at the Max Planck Institute for Polymer research to work on the exploration of chemical space by combining importance sampling and machine learning (ML).

As a theoretical chemist by training, I have witnessed the transformative power of ML techniques on quantum chemistry firsthand, which had a profound impact on my research interests. Intrigued by the potential of statistical learning algorithms in extracting useful representations from massive datasets, I invested time to learn the nuts and bolts of ML, mainly via extracurricular courses and activities. Despite their successful application on a variety of tasks, neural networks—the workhorses of deep learning—are not flawless but suffer from several shortcomings, e.g., they are resource-hungry, easy-to-fool, overconfident in their predictions, and show weak transferability across domains. For me and many others, these flaws are a great source of inspiration that motivates further research.

I am enthusiastic about advancing our understanding of ML techniques as well as engineering smarter models and algorithms, which will help us to solve scientific problems and tackle societal challenges.

This website serves to document my journey. Maybe people with similar interests will find it useful.