Worked with students on networking strategies, resume reviews, and interview prep for software-centric and data science-related roles

Evaluated RoBERTa base model on word deletion, inflection, synonym substitution and misspelling attack using a novel greedy based attack strategy to achieve successful attack rate in the range of 32-86 %

Wrote unit tests (≥86 % coverage) ensuring TensorFlow layers matched their PyTorch counterparts, and merging the code into a unified codebase with existing continuous integration

First author (joint) on the paper ‘Box Embeddings: An open-source library for representation learning using geometric structures’ [EMNLP 2021]