I lead the UT Systems and Storage Lab. I am also a part of the LASR research group.

Research. My group aims to help build the next generation of storage systems. This means building storage systems that have higher performance and stronger reliability. We build both storage systems and the tools need to develop such systems rigorously, such as testing frameworks. Our work involves innovation both at the data-structure level and at the systems level. We open-source all the software we build, and aim to have impact both inside academia and in industry.

Student Recruiting: I am interested in recruiting one or more PhD students to work on problems at the intersection of systems, storage, and machine learning. I am not currently hiring Masters students. If you are interested in working with me, please check out my page for prospective students and send me an email (after reading this!). I do not look at GRE scores when recruiting students. Letters and research experience are the biggest factors for me. I generally interview students (remotely), and that is the strongest signal of whether I would work with them. Make sure to apply to the PhD program by the deadline (December 15).

Recent major projects (See all projects here):

• LiPT. A framework to train ML models faster by identifying the crucial data items in the dataset and focussing on them. ICML 23. Star
  
• Chipmunk. Tool to test whether persistent-memory file systems recover correctly from crashes. Eurosys 23. Star
• Dinomo. Key-value store for disaggregated persistent memory. VLDB 23. Star
• WineFS. Persistent-memory file system that maintains performance as it gets old and full. SOSP 21. Star
• RainBlock. An IO-efficient public blockchain with 10x higher throughput than Ethereum. ATC 21. Star
• MONet. Framework for reducing memory consumption of DNN training. ICLR 21. Star
• SplitFS. File system for persistent memory which reduces software overhead. SOSP 2019. Star
• RECIPE. Concurrent, crash-consistent persistent-memory indexes. SOSP 2019. Star

Current projects:

• Building a persistent-memory file system in Rust (Hayley LeBlanc)
• Making distributed training of ML models more efficient (Yeonju Ro)

Recent impact:

• Meta uses ideas similar to CheckFreq (FAST 21) to checkpoint their large models during training.
• VMware is releasing SplinterDB (ATC 21) as an open-source key-value store!
• Crashmonkey-generated tests have been added to xfstests, the file-system test suite for the Linux kernel (msg, patch)
• Crashmonkey found 10 long-standing bugs in btrfs and F2FS in the Linux kernel.
• Crashmonkey found a bug in the FSCQ verified file system which caused the file system to not persist data on fdatasync. The bug was in the unverified part of the file system (Haskell-C bindings).

The research in our group has been supported by the NSF, the BigHPC consortium, VMware, Google, and Meta.

From 2016 to 2022, I was an Affiliated Researcher at VMware Research.

Teaching

Aug 2023: I am back from sabbatical, and back to research and teaching at UT!

Mar 2023: I wrote a book, The CS Assistant Professor Handbook, for new and aspiring assistant professors tht describes what the professor job is like. Take a look!

Aug 2022: I will be on sabbatical at VMware Research in Fall 22, and EPFL in Spring 23.

Sep 2021: I will be serving as one of the general co-chairs for SOSP 21.

Mar 2021: I co-founded a new diamond open-access journal, the Journal of Systems Research. Check out this blog post explaining its goals and design.

Jan 2021: Congratulations to my student SeKwon Lee for being awarded a 2021 Microsoft Research PhD Fellowship! This is the second student in my group to win this award, after Jayashree Mohan in 2019!

Tigon: A Distributed Database for a CXL Pod
Yibo Huang, Haowei Chen, Newton Ni, Vijay Chidambaram, Dixin Tang, Emmett Witchel
To appear at OSDI 25