Hao Kang

1st-year CS PhD student · Carnegie Mellon

Co-Advised by Chenyan Xiong & Tianqi Chen

Research on agent-native systems for LLM training and hardware-aligned architectures for foundation models.

ScholarGitHubLinkedIn
Portrait of Hao Kang
Deep down my heart, I love writing code that makes things happen!

Selected Publications

  1. Under Review

    Delta-Matching: Closing the Final Gap of Native 8-bit Training for LLMs

    Haozhan Tang, Hao Kang, Han Cai, Song Han, Chenyan Xiong

    Code

  2. NeurIPS 2026

    PithTrain: A Compact and Agent-Native MoE Training System

    Ruihang Lai*, Hao Kang*, Haozhan Tang, Akaash R. Parthasarathy, Zichun Yu, Junru Shao, Todd C. Mowry, Chenyan Xiong, Tianqi Chen

    CodeBlog

  3. NeurIPS 2026

    Benchmark Test-Time Scaling of General LLM Agents

    Xiaochuan Li, Ryan Ming, Pranav Setlur, Abhijay Paladugu, Andy Tang, Hao Kang, Shuai Shao, Rong Jin, Chenyan Xiong

    CodeBlog

  4. EMNLP 2025

    ResearchArena: Benchmarking LLMs' Ability to Collect and Organize Information as Research Agents

    Hao Kang, Chenyan Xiong

  5. arXiv 2025

    FLAME-MoE: A Transparent End-to-End Research Platform for Mixture-of-Experts Language Models

    Hao Kang*, Zichun Yu*, Chenyan Xiong

  6. NAACL 2025

    Interpret and Control Dense Retrieval with Sparse Latent Features

    Hao Kang, Tevin Wang, Chenyan Xiong

* Equal contribution

About

My CS journey all started with a Python tutorial on YouTube. It was immediately clear after the first few hours of watching that once I learned how to program, I could build anything. The pursuit of that kind of freedom has carried me through my undergrad years at CMU and into the work I do today. Never did I anticipate back then how the power of AI would amplify that momentum.

I believe that technology, when guided by an innovative mind, can make a positive impact on this world. To this end, I want to write the code that makes things happen :-)— Hao