About Me

I recently graduated with an M.S. in Computer Science from Purdue University, where I previously pursued a Ph.D. advised by Prof. Chunyi Peng. Before that, I was advised by Prof. Xuehai Qian for one and a half years. Before joining Purdue, I received my B.E. in Automation from Tsinghua University.

My work focuses on large-scale ML / LLM training and serving, distributed training systems (FSDP, ZeRO, data/model parallelism), training and inference efficiency, GPU/CPU memory optimization, and performance profiling for production-scale AI workloads. I have built ML infrastructure across both industry and research, including internships at ByteDance Seed and Meta Ads Infra. I enjoy building scalable ML systems and solving challenging performance bottlenecks in real-world environments.

Open to Opportunities

I am actively exploring full-time opportunities starting Summer/Fall 2026.

Target roles: ML Infrastructure / Platform Engineer · LLM Training & Inference Engineer · AI/ML Systems Engineer

Location: Bay Area and Seattle preferred; open to other U.S. locations.

If your team is hiring for AI infrastructure, ML systems, or LLM training/serving roles, I’d be happy to connect — please reach out via email.

News

• [7/2026] Starting Software Engineering Internship at Waymo – AI Training Infrastructure.
• [5/2025] Started Research Scientist internship at Meta Platforms – Ads Training Infra, Sunnyvale, CA, working on Distributed ML Systems.
• [11/20/2024] Jiaxin Du and I received MobiCom 2024 Best Poster Award.
• [11/1/2024] Our short paper “D-AirPatrol: A Dual-Layer Architecture for Traffic Patrol From the Sky” has been accepted by MobiCom’24.
• [5/2024] Started Research Scientist internship at ByteDance – Seed, Bellevue, WA, working on LLM Infrastructure.

Experience

• Waymo
  Software Engineering Intern, AI Training Infrastructure · Starting Jul 2026

• Meta Platforms – Ads Training Infra, Sunnyvale, CA
  Research Scientist Intern in Distributed ML Systems (Mentor: Shuai Yang), May 2025 – Aug 2025
  • Collaborated with the FAIR PyTorch/FSDP team to co-design and implement an Automatic Selective Unsharding framework atop SimpleFSDP (a PyTorch 2.0 compile-friendly FSDP), improving communication–computation overlap and training efficiency.
  • Built a hierarchical memory profiler and adaptive unsharding algorithm leveraging GPU memory budget for selective parameter retention, achieving up to 5.5% QPS improvement on 8×GPU ads-model benchmarks.
  • Enhanced SimpleFSDP infrastructure with configurable communication scheduling and integrated memory–computation optimization, enabling future adaptive distributed-training pipelines within Meta Ads Infra.
• ByteDance – Seed, Bellevue, WA
  Research Scientist Intern in LLM Infrastructure (Mentor: Yanghua Peng), May 2024 – Sep 2024
  • Supported an LLM training simulator to profile and simulate the training latency of LLMs (especially diffusion transformer models) on 1k–10k GPUs for text-to-video generation and customized computing kernels to achieve 95%+ accuracy.
  • Supported memory profiling and distributed training simulation in frameworks like torch.distributed and DeepSpeed-Megatron within the LLM training simulator and achieved 95% accuracy.
  • Supported PyTorch FSDP and various parallelism techniques in ByteDance’s open-source distributed training stack.
• Purdue University
  • Teaching Assistant
    • CSCI 48700 Artificial Intelligence (2025 Fall)
    • CS 18200 Foundations Of Computer Science (2025 Spring)
    • CS 25000 Computer Architecture (2024 Spring)
    • CS 53600 Data Communication and Computer Networks (2023 Fall & 2024 Fall)
  • Research Assistant in Dr. Xuehai Qian’s Lab (Aug 2022 - Sep 2023)

Portfolio

Please go to the portfolio page: Portfolio Please find my portfolio here: pdf link.

Awards

• MobiCom’24 Best Poster Award; Oct 2024
• Technological Innovation Scholarship of Tsinghua University; Oct 2021
• 2019 & 2020 Hage Foundation Scholarship; Apr 2019 & Apr 2020
• AI Competition of Tsinghua University Third Award; Jun 2021
• Winning Prize of Electronics Design Contest of Tsinghua University Aug 2020