原文 · 未翻译
Announcing the Recipient of the 2026 LMSYS PhD Fellowship - LMSYS Blog | LMSYS Org
[](https://www.lmsys.org/) Projects Blog About Donations Contact
[](mailto:contact@lmsys.org "Email")[](https://twitter.com/lmsysorg "X / Twitter")[](https://slack.sglang.io/ "Slack")[](https://youtube.com/@lmsys-org "YouTube") Projects Blog About Donations Contact
[](mailto:contact@lmsys.org "Email")[](https://twitter.com/lmsysorg "X / Twitter")[](https://slack.sglang.io/ "Slack")[](https://youtube.com/@lmsys-org "YouTube")
‹ Back to Blog
‹ Back to Blog Announcing the Recipient of the 2026 LMSYS PhD Fellowship
LMSYS Board June 08, 2026
We are delighted to announce the first recipient of the LMSYS Fellowship Program: Will Lin.
Following the launch of our Fellowship Program and careful review of applications, we selected Will for his exceptional and sustained contributions to open-source AI infrastructure. The fellowship awards up to $50,000 to support his tuition and fees.
Please join us in congratulating Will!
Will Lin is a 6th-year PhD student in the Hao AI Lab at UC San Diego, advised by Hao Zhang. He leads FastVideo, an open-source framework for accelerated diffusion generation and post-training that has grown to 3.7k+ GitHub stars, with integration as a backend in NVIDIA Dynamo. Its modular pipeline architecture has influenced multimodal generation designs in projects including SGLang.
His work spans the full stack of generative video and world model systems: from sparse attention optimizations and quantization to distillation, post-training, and real-time inference. He co-led the first interactive-latency 1080p text/image-to-audio-video generation pipeline, and his recent work includes training pipelines for autoregressive world models.
Will has interned at Anyscale and is currently interning at NVIDIA, where he helps lead and grow FastVideo's collaborations across industry partners and downstream adopters. He began his PhD working on operating systems and datacenter networking with FPGAs before transitioning into machine learning systems three years ago.