Experience

  1. Software Engineer

    Google DeepMind
    • Gemini Spark ✨: Drove the 0-to-1 development of the 24/7 personal AI agent and co-led the Agent Skills track for its Google I/O 2026 launch.
    • Gemini Agent 🚀: Engineered end-to-end multi-step agent planning and latency optimizations to support the product launch.
    • Skills for all: Co-leading the development of shared Agent Skills and Agent Planning infrastructure across all Gemini surfaces (Chat, macOS, and Chrome).

  2. Software Engineer

    Google
    • Core Labs (2024-2025):
    • CoreID Tech Lead (2020-2024):
      • Led a 10-engineer team designing and deploying secure, scalable identity and access management (IAM) services across Google.

  3. Software Engineer Intern

    Google
    • Modeling and prediction of financial time series data via ML

  4. Teaching Assistant & Research Assistant

    Georgia Tech
    • CS 4400 Introduction to Database Systems (Spring/Fall 2019)
    • CS 8803 DML: Data Management and Machine Learning (Fall 2018)
    • Research on Approximate ML via Importance Sampling
    • Research on Automated Feature Engineering: Third Prize in KDD Cup 2019 AutoML Track

  5. Research Intern

    SenseTime
    • R&D on large-scale ML system

Education

  1. MSc in Computer Science

    Georgia Institute of Technology
    • Grade: 4.0/4.0
    • Specialization: Computing Systems
    • Advisors: Prof. Xu Chu

  2. BSc in Computer Science

    Chinese University of Hong Kong
Skills & Hobbies
Technical Skills
Programming Languages

Java/Kotlin, C++, Python, Go, SQL

GenAI & Blockchain

AI Agent, Blockchain, IAM, Micro-services

Distributed Systems

Distributed System, Testing, Project Management

Machine Learning

Deep Learning, Ensemble Methods, ML Systems

Awards
KDD Cup AutoML Challenge
KDD Cup 2019 ∙ August 2019
3rd prize worldwide in developing AutoML solution for temporal relational data classification with a team of researchers from Alibaba Group and Georgia Tech
FlexPS: Flexible Parallelism Control in Parameter Server Architecture
VLDB 2018 ∙ January 2018
  • Built FlexPS, a Parameter Server system with flexible parallelism control
  • Achieved 400% speedup vs Spark MLlib in k-means++ implementation
  • Optimized sparse data communication, reducing server time by 45%
Languages
100%
English
100%
Chinese