CV

Education

• B.S. in Computer Science, University of California San Diego, 2021
  • GPA: 3.918/4.000
  • Honor: Magna Cum Laude

Field of interests

• Computer Systems
• Computer Architecture
• Accelerator
• FPGA/GPU/CPU
• Physics Simulation

Research experience

• January 2021 - July 2022: Research Assistant

  • Project: suturing simulation
  • UCSD Advanced Robotics and Controls Lab
  • Duties included:
    • Created a physics engine for simulating automated surgery using the position based dynamics (PBD) algorithm.
    • Implemented the key components including physics simulator, controller, and data types following best OOP practices using OpenGL and C/C++, as well as OpenMP for parallel computing.
    • On top of the simulation engine, designed a differentiable PBD constraint formula that simulates the movements of particles in the specific case of needle injection.
    • Successfully simulated the deformation of human tissue caused by injecting the needle, proving both the engine and the formula works.
    • Utilizing the gradients, implemented a deep learning controller (Python) that enabled the DaVinci robot arm to suture wounds automatically.
  • Supervisor: Professor Michael Yip

• December 2020 - July 2022: Research Assistant

  • Project: reinventing experience replay
  • UCSD Mattar Lab
  • Duties included:
    • Improved the performance of DQN, a decision-making algorithm by designing novel methods of experience replay.
    • Designed new data storage and retrieving methods for experience replay in Q-learning, inspired by a pattern in human memory replay called “gain and need”.
    • Incorporated this improvement of sampling order to the Prioritized Experience Replay (PER) algorithm.
    • Developed pytorch scripts to train the PER algorithm on multiple GPUs using Kubernetes and Docker.
    • Achieved at most 500% higher scores than humans in playing Atari games (e.g., Battlezone, Asterix).
  • Supervisor: Professor Marcelo Mattar

• September 2020 - June 2021: Research Assistant

  • Project: 3D Bioprinter Software
  • UCSD Chen Lab
  • Duties included:
    • Created a new control software for the 3D bioprinter, enabling researchers to print complex biological tissues layer by layer.
  • Supervisor: Professor Shaochen Chen

Work experience

• Jan 2024 - Present: TikTok inc.

  • Backend Software Engineer
  • Los Angeles, CA
  • Duties included:
    • Working on the Global Monetization team to develop backend services and data pipelines for ad delivery.
    • Maintained and refactored a C++ simulation and forecasting engine, resulting in 50% latency improvement.
    • Created a new forecasting model for ads delivery results which brings down prediction error by 15%.
    • Created data jobs using Spark/Java, supporting weekly production of 60M+ ad targeting data.
    • Optimized the access pattern to the user data in memory, improving cache hit rate and reducing latency by 28%.
    • Designed and built a data collection pipeline using Kafka to support quick profiling of the simulation engine.

• July 2022 - Jan 2023: Qualcomm

  • Camera Software Engineer
  • San Diego, CA
  • Duties included:
    • Designed and implemented camera software and algorithms for various Automotive (car) Customers worldwide.
    • Developed multiple API pipelines in IL (Independent Layer) between application and hardware drivers using C/C++.
    • Maintained the robust operation of 10+ Qualcomm Automotive platforms that allows fast camera dev/testing.
    • Created an automatic Python test module for features across different car sensors, reducing 90% manual work.
    • Followed the full software development cycle and the automotive industry standards (ASPICE).
    • Feature owner of the dynamic sensor orientation, enabling customers to rotate video output of surrounding view sensors.

• Jun 2019 - Aug 2019: Posse.io

  • Software Engineer Intern
  • Encinitas, CA
  • Duties included:
    • Full-stack developer for Buztubr (now Posse.io), the world’s first digital audience monetization trade floor.
    • Developed the registration interface for users with React and HTML/CSS, enabling one-click login functionality.
    • Refactored the old, static product landing page into fully animated pages using various JavaScript modules.
    • Simplified the payment methods registration system and removed 80% redundant code with Stripe API.
    • Created a customizable data board using Facebook API, allowing influencers to view audience statistics in real-time.

Skills

• Programming
  • Python
  • C++
  • C#
• Tools
  • tensorflow
  • pytorch
  • OpenGL