I work full time for Google as Research Scientist.
At Google, I'm a proud member of the XLS team.
I'm also an Associate Professor at the Computer Science department of Northwestern University where I run the
research lab.
At Northwestern University I am a member of the Systems and Networking group, the Programming Languages group, and the Computer Science and Computer Engineering Departments.
My group at Northwestern is passionate about understanding how abstractions used within and around compilers should evolve to better support hardware and applications trends.
This goal often leads us to co-design compilers with the computer architecture and operating system they target as well as with the programming language they translate.
The full list of news can be found here
Research: Congratulations to Tommy, David, and the rest of the team for having a new paper accepted at ASPLOS 2025!
This paper describes a new approach to save bits for tiny devices by safely speculating on the actual bitwidth needs of program's integer variables.
This work was a massive effort from all of us that required almost 6 consecutive years of focus.
This work spans architecture simulator extensions (Gem5), ARM ISA extensions, new code and data profilers, new LLVM-based compilers, new NOELLE-based profiler-guided code analyses and transformations, new energy models, and new gate-level simulators.
This work also includes the integration with other fundamental software infrastructures including DRAMSim and SimPoint.
The entire work has been open-sourced with the hope others can benefit from it.
Career:
What I'm doing at Google is too exciting to stop working on it, the work has great potential (stay tuned for papers), and the team I'm in is just fantastic!
Therefore, I've extended my sabbatical until the end of August 2026.
Research: Congratulations to Yian and the rest of the team for having their new paper accepted at CGO 2026!
This paper introduces the new abstraction, PS-PDG, which extends the Program Dependence Graph to effectively capture the parallel semantics of OpenMP, excluding aspects related to hardware accelerators.
This will enable OpenMP compilers to optimize and adapt the parallelization plan encoded by OpenMP developers to the target platform.
Research: Congratulations to Tommy for having his new paper accepted at CGO 2026!
The paper automates a crucial optimization for data collections -- data enumeration -- which had to be performed manually until now.
This optimization is enabled by MemOIR.
Back to top
Share this page with: