As a data scientist, my work has involved both the development and implementation of end-to-end models for various solutions regarding client operations. In the process, I generally use a variety of techniques, both classical and novel, involved in machine learning. A large part of my work goes into feature engineering prior to actual modelling; after all, the features make up half the technique.
My research centred on the development of inference models and algorithms for applied optimization problems, especially those coming from the field of structural genomics. These problems focussed on cancer genomics, and involved formulating statistical inference models for classifying sequence data from patients. I mainly worked on optimisation algorithms for these problems, as well as the implementation of these ideas and evaluating them on real data.
I have also worked on convex relaxations for certain problems on hypergraphs, the computational complexity of these algorithms, and combinatorial optimisation. I maintain a strong interest in combinatorics and analysis.