Work
My research and engineering experiences
ByteDance (Douyin), Creation & Music Recommendation Team
Research Intern
- Selected for ByteDance's Jindouyun Talent Program (ByteIntern track), a selective recruiting program for high-potential technical talent
- Worked on generative recommendation and LLM4Rec for personalized content generation and discovery
- Applied ideas from CAKE to LLM-driven Bayesian optimization for recommendation and adaptive content selection from feedback
Dria
Research Intern
- Researched evolutionary coding agents for algorithm discovery and codebase optimization
- Built EvolveBench, a benchmark and execution harness for evaluating coding agents on real GitHub repositories
- Co-developed Kai, an autonomous coding agent for vulnerability discovery, exploit verification, and patch generation
- Built Kai's sandboxed execution environments with Docker, E2B, and Vercel Sandbox
- Shipped Kai as a CLI, FastAPI web UI, and chat-platform integrations
Huawei
Research Intern
- Researched scalable black-box optimization for robust parameter tuning in high-dimensional 5G systems
- Developed ZAP, a gradient-free optimizer for high-dimensional Gaussian process training
- Presented the research outcomes as an oral paper at IEEE ICASSP 2026
Bayesian Learning for Signal Processing (BLSP) Group
Undergraduate Research Assistant
- Developed a grid spectral mixture kernel and SLIM-KL, a communication-efficient distributed GP kernel-learning method based on quantized ADMM
- Published the research outcomes at IEEE FUSION 2022 and IEEE TNNLS 2025
Shenzhen Research Institute of Big Data (SRIBD)
Undergraduate Research Assistant
- Investigated federated matrix factorization for data clustering and recommender systems
- Developed FedMAvg, which combines alternating minimization with model averaging for federated matrix factorization
- Presented the research outcomes as a poster at IEEE ICASSP 2021