Experience

• Develop classification, sequence-labeling, span-prediction, and information extraction systems with Python, PyTorch, Transformers, CRFsuite, and scikit-learn.

• Built hierarchical and PPO-based approaches for class-imbalanced negation detection and related biomedical NLP analysis.

• Designed sparse and dense retrieval pipelines with BM25, Pyserini/Lucene, MedCPT, FAISS, cross-encoder reranking, and TF-IDF/MMR.

• Processed 26.8 million PubMed documents and built evaluation workflows for answer accuracy, retrieval quality, evidence grounding, and citation reliability.

• Supported COMP 479 (Information Retrieval and Web Search) and COMP 232 (Discrete Mathematics) through tutorials, office hours, grading, and assignment guidance.

• Helped undergraduate and graduate students translate core information retrieval and discrete mathematics concepts into working solutions.

• Provided technical clarification on assignments, evaluation criteria, and project implementation details.

• Developed application features and workflow automation for Kissflow Procurement Cloud across purchasing, approvals, supplier management, and source-to-pay flows.

• Designed workflow rules, forms, validations, and approval paths from business requirements in Kissflow's low-code platform.

• Built API services on Google Cloud Platform and partnered with backend and product teams to integrate procurement data and ship tested features.

• Developed a real-time esports broadcast overlay using Python and Node.js.

• Built backend logic that updated match and event counters dynamically during live streams.