Om Shah · San José, California
Applied mathematics,
built like working software.
Applied Mathematics student at SJSU, undergraduate researcher in geometric topology, and a robotics hacker who ships autonomous systems. I care about proofs that hold and code that deploys.
Research
B.S. Computational & Applied Mathematics
San José State University · transferred from Moorpark College
Student Assistant - Geometric Topology
At the San José State University Research Foundation, working with Assistant Professor Edgar A. Bering IV on a novel generalization of Theorem 5.1 of Coloring the Curve Graph (Gaster, Greene, Vlamis, 2016) from the curve graph of a surface to the sphere graph of a connected sum of copies of .
The curve graph admits a proper coloring whose chromatic number is controlled by the topological complexity of the surface. We ask the analogous question for the sphere graph of the manifold :
The bound is realized by a coloring built from the geometric intersection number, mirroring the surface case where Gaster, Greene, and Vlamis establish
What I build
Two threads, one habit: take a hard system, find the right abstraction, and make it dependable. Switch between them below.
Athyna
Omm2005/AthynaAn open-source AI research companion that helps you collect, organize, and interrogate papers - keeping the reading loop tight so the thinking stays the hard part. Built with a typed, end-to-end stack and a schema-first data layer.
- Next.js
- TypeScript
- Tailwind
- Supabase
- Drizzle ORM
- Type safety
- end-to-end
- Data layer
- schema-first
- License
- open source
Autonomous Robotics
Moorpark CollegeRobotics Developer and tutor building autonomous competition systems that reached the World Championships. The control stack closes the loop with a PID controller over wheel odometry, lifting positioning accuracy by 0%.
The control law
Engineering practices
CI/CD by default
Every change runs through automated build, test, and lint gates before it merges. Green pipeline or it does not ship.
Measured, not guessed
Decisions follow metrics - the 150% robotics accuracy gain came from instrumenting the loop and tuning against the numbers.
Accessible & predictable
Keyboard-first navigation, visible focus, semantic structure, and layouts that behave the same on every screen.
Typed end to end
Strong types and schema-first data from the database to the UI, so refactors are safe and the compiler catches the rest.
Outside the proofs
Queen, on repeat
'Bohemian Rhapsody' is the closest thing I have to a north star - six minutes that change key four times and still resolve. Ambition with structure.
High-fidelity listening
Lossless tracks, a quiet room, and good gear. Hearing the detail in a mix is its own kind of close reading.
Gymmaxxing
A five-day split, run with the same discipline as a proof - progressive, measured, and consistent over flashy.
Riding the rails
Exploring Southern California by public transit and train. Slower than driving, better for watching a system actually move people.
Contact
Open to research collaborations and hard engineering problems. The code lives on GitHub - the fastest way to see how I work is to read it.
Built with proofs in mind.