Experience
2023 — Now
2023 — Now
Milpitas, California, United States
Owned the development of Motional’s next-generation data mining system from inception to deployment:
• Transformed vision into robust system & software design, securing leadership buy-in
• Implemented the Python-based mining system with features like code-free configuration (Hydra), efficient & robust data ingestion (AWS S3, PyArrow, Pandera), workflow orchestration (Airflow, Flyte), and parallelized processing (NumPy, Pandas, etc.), achieving a 10x runtime increase
• Reduced annotation costs by 80% with innovative, ground truth-free data mining approaches
• Cut runtime for terabyte-level workloads from days to hours by scaling with distributed compute (Ray on k8s)
• Fostered company-wide collaboration across multiple research teams, mining over 50k scenes to date
2021 — 2023
2021 — 2023
San Diego, California, United States
• Developed a code-free scenario mining tool for complex queries on petabyte-scale data using Looker, LookML, and SQL on top of Trino, empowering a 4x larger, non-technical user base
• Optimized query speed (5x increase) using Looker’s Persistent Derived Tables (PDTs) and query caching
• Led R&D for TuSimple’s first data storage optimization initiative, achieving ~$600k in annual cost savings
• Built an optimization pipeline processing terabytes of data daily using Python, Trino, AWS S3, and Airflow, leveraging strategies formulated from data modeling with 1B+ user activity records
• Created an interactive org chart with 5000+ monthly page views using JavaScript as a side project
2020 — 2021
2020 — 2021
San Diego, California, United States
PrecisePK is the leading therapeutic drug monitoring software provider trusted by clients globally.
Initiated & led the development of product localization, boosting new leads overseas by ~75%:
• Engineered full-stack using React.js & C++; introduced Scrum and facilitated sprints
2017 — 2019
2017 — 2019
La Jolla, CA
• TA for multiple classes, covering topics such as Java, C, C++, object-oriented programming (OOP), advanced data structures, computer organization and system programming, and advanced search/reasoning/reinforcement learning algorithms
• Held lab hours to analyze and debug students’ code as well as explain related concepts
• Led discussion/review sessions for 400+ students, allowing them to understand their coursework further
2019 — 2019
2019 — 2019
San Francisco, CA
Worked on the Android App for Salesforce’s field service management system:
• Built the TalkBack feature in Java & Kotlin, doubling task success rate for vision-impaired users
• Integrated Kotlin code coverage into the CI pipeline, lowering manual testing costs by 20%
• Decreased task time by 25% with the implementation of the WebView-to-PDF feature using Java
Education
UC San Diego