Skip to content
View halfrost's full-sized avatar
๐Ÿ‘€
ๅพฎไฟกๅ…ฌไผ—ๅท๏ผšไบ”ๅˆ†้€‰ๆ‰‹
๐Ÿ‘€
ๅพฎไฟกๅ…ฌไผ—ๅท๏ผšไบ”ๅˆ†้€‰ๆ‰‹

Sponsors

@xiaomaimai

Block or report halfrost

Report abuse

Contact GitHub support about this userโ€™s behavior. Learn more about reporting abuse.

Report abuse
halfrost/README.md

Machine Learning Systems, Alignment, and Evaluation

I am a research-oriented machine learning systems engineer working on foundation model infrastructure, alignment, and evaluation. I build efficient and reliable systems for large language models while studying the algorithms and data choices that make them more useful, controllable, and cost-effective in real applications.

  • ๐Ÿง My central research interest is model-system co-design: understanding how model architecture, inference algorithms, data curation, hardware utilization, scheduling, and distributed runtimes interact.
  • ๐Ÿ’ผ At TikTok, I work on Model-as-a-Service platforms and high-performance LLM inference, developing production serving infrastructure with vLLM and SGLang.
  • ๐ŸŽ“ My recent research includes distributed disaggregated inference, preference optimization, instruction-tuning data selection, multimodal evaluation, and retrieval-augmented biomedical summarization.
  • ๐ŸŒฑ I investigate alignment and evaluation methods that connect measurable model behavior with real-world usefulness, controllability, reliability, and serving cost.
  • ๐Ÿ“š My systems work spans model runtime integration, scheduling and continuous batching, KV-cache and memory management, distributed execution, observability, and reliability.
  • ๐Ÿ’ป My broader research experience includes reinforcement learning for robotics, healthcare sequence modeling, privacy-preserving machine learning, and motion planning.
  • โ›ต I am interested in collaborating on open research and infrastructure that make frontier AI systems faster to experiment with, more rigorous to evaluate, and dependable at scale.
  • โœ๐Ÿป I share technical writing on machine learning systems, infrastructure, and software engineering through my personal blog.
Some other achievements about me~e~e
  • ๐Ÿ’™๐Ÿ’› Be proud of the University of California, Berkeley. ๐Ÿป Proud California Golden Bear. Fiat Lux โœจ Go Bears.
  • ๐ŸŒฒ Be proud of Stanford University. โค๏ธ Proud Stanford Cardinal. Die Luft der Freiheit weht.
  • ๐Ÿงฃ Be proud of Carnegie Mellon University. ๐Ÿพ Proud Carnegie Mellon Tartan. My heart is in the work.
  • ๐ŸŽ‰ Professional Membership of ACM / IEEE / IEEE-CS / CCF / Sigma Xi.
  • ๐ŸŽ Apple Developer.๐Ÿ‘จ๐Ÿปโ€๐Ÿ’ป & Apple Teacher.๐Ÿคช

  • ๐Ÿ“Š Open-source activity and repository highlights:

halfrost's Github Stats halfrost's Github Trophy


Explore my repositories, research interests, and technical writing, or reach out to discuss machine learning systems and frontier AI.

visitor badge


Pinned Loading

  1. kubernetes/kubernetes kubernetes/kubernetes Public

    Production-Grade Container Scheduling and Management

    Go 123k 43.4k

  2. golang/go golang/go Public

    The Go programming language

    Go 135k 19k

  3. Halfrost-Field Halfrost-Field Public

    โœ๐Ÿป ่ฟ™้‡Œๆ˜ฏๅ†™ๅšๅฎข็š„ๅœฐๆ–น โ€”โ€” Halfrost-Field ๅ†ฐ้œœไน‹ๅœฐ

    Go 13.2k 1.9k

  4. LeetCode-Go LeetCode-Go Public

    โœ… Solutions to LeetCode by Go, 100% test coverage, runtime beats 100% / LeetCode ้ข˜่งฃ

    HTML 33.8k 5.7k

  5. threes-ai threes-ai Public

    ๐Ÿ† Deep Reinforcement Learning for the Threes! game.

    Go 163 39

  6. vllm-project/vllm vllm-project/vllm Public

    A high-throughput and memory-efficient inference and serving engine for LLMs

    Python 85k 18.8k