Divyesh Bommana

• Contributing to Professor Tianyu Jiang’s NLP research project focused on recognizing embodied emotions in natural language. • Developed scraping techniques using Python to collect and preprocess data from Reddit meeting specific linguistic criteria while addressing challenges like API limitations and data quality. • Submitted a paper to the Association for Computational Linguistics (ACL) for review, extending upon previous datasets and showcasing embodied emotion detection and classification. • Currently analyzing data to research on emotion recognition and explore the integration of Multi-Task Learning models to include audio or visual inputs.

• Built Natcast’s first R&D cloud environment on Microsoft Azure, contributing to the organization’s role in the National Semiconductor Technology Center (NSTC), a NIST initiative advancing U.S. semiconductor R&D. • Installed, validated, and extended the Lithography Benchmark (LithoBench) Optical Proximity Correction (OPC) codebase while advancing two core ML tasks - lithography simulation and mask optimization - through experiments with state-of-the-art models such as DOINN and GAN-OPC. • Modified physics-based lithography simulations to investigate parameter sensitivity, including research into novel optical kernels from academic literature. • Conducted exploratory research on per-shape detection and analysis of printed wafer patterns, identifying geometric features where models underperformed.

• Developing and deploying an incentive-driven maternal health app at Nationwide Children's Hospital and The Ohio State Wexner Medical Center to combat infant mortality. • Leveraging AWS Amplify and React Native to create scalable, user-friendly solutions. • Key responsibilities include reviewing and optimizing the codebase, feature prototyping, implementing solutions for enhanced usability, data handling and creating python scripts to gather back-end metrics of users for further research studies.

Worked with Green and Associates to develop a real estate-based web application using the React framework. This application was tailored for individuals in the early stages of buying and selling a home to help guide them through pre-stages and provide them with accurate and up-to-date information. This platform was designed to contain features such as current steps, credit score updates, home searches, tips and tricks, FAQs, direct messages, and AI support.

Under the guidance of Professor Golrokh Mirzaei, I created an effective Multi-Task Learning model that could identify three types of cancer (breast, colorectal, and pancreatic) and their subtypes using mRNA sequence data. I developed this model in Python with Jupyter Notebooks and the TensorFlow library, using training data from cBioPortal. I presented my research at a Scholar’s seminar and showcased a poster at the Undergraduate Research Colloquium at OSU Marion, titled “Multiomics-driven Multi-task Learning for Cancer Classification.”