Nikil Shyamsunder । निखिल् श्याम्सुन्दर् ।
I study computer science and mathematics at Cornell University. I am broadly interested in formal methods, high-performance computing, compilers, and computer architecture.
I work on Protocols, a DSL for specifying the communication of digital circuits, in the Computer Architecture and Programming Abstractions Lab (Capra) under Adrian Sampson. In Spring 2026, I was at AWS Annapurna Labs working on formal verification and graph-level optimizations for the Neuron compiler for Trainium.
I enjoy hiking, snowboarding, classical languages (Latin and Sanskrit), and recently, Victorian literature.
News
| Aug 04, 2026 | Our paper (from my time at AWS) on “A Barrier-Free Synchronization Algorithm for Multi-Engine AI Accelerators” was accepted to CGO ‘27. |
|---|---|
| Jul 22, 2026 | My work on “Formal Performance and Compile Time Guarantees for Compiler Optimization Heuristics” was accepted to the FMCAD ‘26 Student Forum. |
| May 06, 2026 | I gave an invited talk at UC Berkeley’s Slice Lab on our Protocols DSL. |
| Mar 05, 2026 | I was inducted into Cornell’s chapter of the Phi Beta Kappa Honor Society. |
| Dec 16, 2025 | Cynthia and I published a blog post on building a backend for Allo accelerator design DSL to Google’s XLS for ASIC synthesis. |
| Oct 10, 2025 | Our work on the Protocols DSL was accepted to the FMCAD ’25 Student Forum, where I presented in October. |
Research
- A Barrier-Free Synchronization Algorithm for Multi-Engine AI Accelerators
Chungha Sung*, Nikil V. Shyamsunder*, Hanliang Zhang, Daniel Kroening, and Joonwon Choi
To appear in CGO 2027
*Equally contributing authors - Specifying Hardware Communication as Programs
Ernest Ng, Nikil Shyamsunder, Francis Pham, Adrian Sampson, Kevin Laeufer
[arXiv PDF] [Code] - Making Machine Learning-based Snow Water Equivalent Forecasting Research Productive and Reusable by Geoweaver
Sun, Z., Cristea, N. C., ... Shyamsunder, N., ... Bhandaru, P.
American Geophysical Union, Fall 2022
Talks
- Formal Performance and Compile Time Guarantees for Compiler Optimization Heuristics
Student Forum Talk and Poster, FMCAD '26, Sep 2026 - Specifying Hardware Communication as Programs
Invited Talk, UC Berkeley EECS SLICE Lab, May 2026 [Slides PDF] - Building an Interpreter for an Imperative Hardware Interface Specification Language
Student Forum Talk and Poster, FMCAD '25, Oct 2025 [Poster PDF]
Experience
- AWS Annapurna Labs (Jan 2026 – Present)
Formal verification of compiler passes in Lean 4 and graph-level optimizations for the Neuron compiler. - Lumen Technologies (May 2025 – Aug 2025)
Engineered scalable AWS-based data pipelines and dashboards in Angular to process and visualize real-time network threat intelligence for millions of IPs. - Beacon Platform (May 2024 – Aug 2024)
Built a full-stack testing dashboard and root-cause analysis system, reducing regression failures by 22%.
Teaching
- CS 2800: Discrete Mathematics
- Fall 2025 (Head TA)
- Spring 2025 [Induction on Trees Video]
- Fall 2024
Writing
| Nov 07, 2025 | Optimizer vs. Learner: Strategic Manipulation in Games and Auctions |
|---|---|
| Jul 14, 2025 | Using Hadoop to Analyze 32 Million Movie Ratings |
| Aug 08, 2024 | A Case Against Supreme Court Term Limits |