I'm a 2025 graduate from Tufts University with a B.S. in Computer Science. I'm interested in systems programming, including networks, operating systems, compilers, and parallel computing.
I've recently become very interested in the software infrastructure that powers Machine Learning workloads.
This project is a high-performance reimplementation the Universal Machine from CS40, the Machine Structure and Assembly Language Programming course at Tufts. The program JIT compiles Universal Machine assembly programs to native machine code and executes it. This alone achieves a 3x performance improvement over a emulation-based approach (see GitHub for more details).
The Universal Machine also uses a custom memory allocator that I built and presented at JumboHack 2025 with my hackathon teammates. This allocator cut down on the JIT compiler runtime by an additional 12.4%. The combination of the JIT compiler and memory allocator led to an overall 3.5x performance improvement over the baseline emulator.
To my knowledge, this is the fastest UM implementation that exists. Future Tufts students, I would love to hear from you if you beat my UM. See GitHub for my comprehensive profiling results.
This proxy parses HTTPS requests from the user's browser (or a command line client). All requests are redirected to their intended destination, and most traffic is forwarded through the proxy without being modified. Quora response content is augmented with a fact-checking tool. The user can highlight text that they want fact-checked, and use the tool to forward the text in question to an LLM for accuracy analysis.
Due to the nature of SSL, the proxy must become a certificate authority in order to be trusted by the HTTPS client. This is necessary to decode the encrypted response, inject the JavaScript to include the fact-checking tool, and then re-encode the response and send it back to the client. The proxy was written from scratch in C aside from the OpenSSL libary.
The two projects above are from the Networks course at Tufts. Both have been rigorously tested for reliability with Python/shell clients.
A complete business plan for an open-source tractor software company supporting right-to-repair. Includes 5 years of pro forma financial statements.
The source code for this website. Hosted on AWS EC2 with automatic deployment.
Basketball is my favorite sport to play and watch. I love the way the game rewards great team chemistry. I played guard in middle school and high school, played intramural in college, and still play pickup occasionally. I take pride in efficient shot selection and being an effective facilitator on offense, as well as playing good help defense.
I loved pitching as a kid playing little league baseball, and have recently rediscovered my love for pitching through wiffleball. I've mastered the slider and can consistently throw it for a strike. I also have a very effective riser, but can't reliably throw it for a strike yet. I'm still learning to throw a screwball, sinker, and changeup.
I drive a 2003 BMW 330xi 6MT. I've owned it since January 2024 and have mostly maintained it myself, This has included significant repairs such as the rear driveshaft and starter, as well as routine maintenance like brakes, spark plugs, and oil changes. The car has 222,000 miles and runs strong.
More coming soon....
