Yuwei C.

YouTube Trust and Safety Team, working on LLM output evaluation as a Data Science research intern - Researched and designed evaluation dimensions and metrics for AI-generated metric report summaries, contributing to the automation of the metric reporting process. - Developed a Python library implementing the proposed LLM-based rater with prompting strategies that evoke chain-of-thought reasoning for improved interpretability and diagnostic insights. - Conducted comprehensive sensitivity analyses to evaluate metric fidelity and collected human-written summary samples to align automated ratings with human preferences. - Generated scorecards for AI-produced summaries and provided product improvement recommendations based on the evaluation results.

Modeling COVID-19 Transmission in Mid East Asia Countries Calibrated age-stratified SEIR model parameters based on country-specific surveillance data and summary statistics Added features, for instance, integrating Google mobility data to reconstruct what had happened to the age-stratified model Quantified local intervention intensity based on WHO Public Health Social Measure dataset to explore whether intervention policies could alter the public’s mobility pattern Assessed policy benefits, like social distancing and elderly shielding by model simulation

Conducted literature review on global value chain of semiconductor industries Found Singapore semiconductor industry leaders based on statistics such as revenue, employment, and produced a detailed local industry analysis at the end of internship

Conducted literature reviews on reinforcement learning Established testing dataset by collecting cargo instances online Built statistical models to help improve performance of the cargo loading system by 5%