My Work
Professional Experience
Machine Learning Engineer
Asurion
San Francisco, CA
Oct. 2024 - Present
• Led team of 4 to design GraphRAG pipeline, improving multi-step logical reasoning accuracy by 40% across 1000+ test cases • Pioneered voice AI streaming architecture with OpenAI and Twilio, achieving 4x latency reduction over traditional solutions • Devised and deployed systematic benchmarking system to analyze multimodal LLM performance across 2k+ monthly inferences, identifying cost-optimal configurations for enterprise deployments
Data Scientist
Towngas
Hong Kong
Jun. 2023 - Sept. 2023
• Engineered automated ETL pipeline using PySpark to restructure 300TB unprocessed database • Minimized query delays through database optimization • Built and optimized XGBoost classifier model reducing technician deployment redundancies by 30%
Software Engineer, Data Platform
Sensing and Robotics for Infrastructure Lab
Los Angeles, CA
Nov. 2023 - Jul. 2024
• Designed and deployed a GNN-based Roadwork Prioritization Platform with React and Node.js • Implemented real-time model tuning and feature engineering • Developed 30+ user-customizable importance metrics
Python Developer
The American Academy of Pediatric Neuropsychology
Long Beach, CA
Aug. 2022 - Dec. 2022
• Engineered a headless browser web scraper with Selenium to extract 100,000+ records • Cleaned and wrangled patient records to MySQL DBs for k-means clustering, lowering outreach costs by $1.5k per month • Maintained full documentation in Git to help onboard 20+ business-centric users
Software Engineer, Intern
Change For Change
Los Angeles, CA
Dec. 2021 - Nov. 2022
• Developed a Chrome Extension for an NGO's e-commerce donation platform • Led 4 developers to build a RESTful Twitter bot for non-profit posts • Reached 350 followers within a week • Designed wireframes for user portals in Figma
Personal Projects
Anthropic & Menlo Builder Day
Oct. 2024 - Nov. 2024• Led 3-person team to win 2nd place ($55,000) among 100+ teams • Reinvented Google's captcha-system • Improved standard false-negative rate by 70%
AWS & Slalom Innovation Challenge
Mar. 2023 - May. 2023• Devised and built ML-driven parking solution • Collaborated with AWS and Slalom engineers • Designed solution in Figma for UCLA's campus-wide competition
Research Analyst
Sept. 2022 - Mar. 2023• Collaborated in 5-person team • Developed contextual Wikipedia article recommender model • Implemented solution using PyTorch on WikiCS dataset