(147e) Machine Learning Scientist, Especially for Chemicals, Materials, Multi-Omics, Health, and Environment.

My research focuses on exploring the power of deep learning techniques in problems that are difficult to solve using state-of-the-art experimental or theoretical approaches. I am driven by the applicability of my research towards human and animal health and well-being, environmental concerns of this century, the next generation of materials, and sustainable material development. Because of my industry experience, I envision a part of my work to bridge the gap between industrial and academic research and collaboration, which I believe, is of paramount importance when thinking about the immediate welfare of our planet.

My research has mainly focused on two areas: (1) Machine Learning (ML) pipelines for molecules, macromolecules, and micrographs. Whether it's a problem of establishing structure-property relationships, finding binding sites on proteins, or the responsible base-pair sequence on a strand of a DNA, figuring out molecular hotspots have been an area of immense importance. When constructed properly, I believe that trained ML models can reveal this information. I work on data representation for molecules requiring little to no feature engineering. Similarly, microscopy is a crucial characterization tool for biologists and materials scientists. For the purposes of ML, micrographs present a more complex problem than other image data because they contain additional artefacts/features, sometimes implicitly relevant to the study. Thus, micrographs are another area where I believe ML approaches will be particularly useful. (2) ML for industrial chemical formulations. A successful ML model could be one that predicts light transmittance or thermal conductivity, for instance, and maps it to the oligomers in the formulation and the curing parameters used. I am confident that deep neural networks have immense potential in capturing the complex mechanical-chemical properties of any formulation.

My current research involves building machine learning (ML) based pipelines for detecting airborne microbes, onset of diseases, and chemical discovery algorithms to find suitable ligands for virus purification. Previously, I worked in the area of experimental auxetic metamaterial development for my doctoral and first postdoctoral work. This was an area dominated by theoretical work, where we became pioneers in commodity auxetic materials and designed composites inspired from tensegrity structures. My industry experience was in a fast-paced R&D environment, where I managed a team of 5 scientists, working on functional coatings, formulations, and scale-up operations. I led several lab-to-production trials.