Research Assistant at National University of Singapore

Built a GraphRAG-style retrieval pipeline in Python, combining FAISS vector search with NetworkX graph traversal and reranking, increasing QA relevance by 60% relative to a baseline RAG on an internal evaluation set. Fine-tuned an LLM for domain-specific analysis, evaluated performance with benchmarking workflows, and compared results against alternative models to select the best-performing approach. Automated large-scale data collection using Selenium, reducing data collection time by 80% and enabling repeatable dataset builds for analysis.

Commercial Excellence Intern at Grab

Built a curated SMB performance analytics framework using Python and PySpark to rank high-potential brands for enterprise conversion, enabling Business Analysts and Account Managers to prioritize outreach and maximize revenue opportunities. Built and maintained a cross-vertical datamart by implementing ETL pipelines, improving data architecture, and optimizing downstream query performance to enable self-serve analytics for multiple commercial teams. Partnered with stakeholders across GrabFood, GrabMart, Grab for Business, and Omnichannel to deliver recurring reports and deep-dive analyses; defined KPIs, optimized SQL, and reduced manual reporting effort by 70%.