Computer Science,Filmmaking,Writing

usr/ankit

Ankit M. this side!

I work on the inside of systems and on getting other people in there with me.

Operating systems, language models, digital circuits, software development, screenplay. I am finishing a BS and an MS in Computer Science at the University of San Francisco, with AI concentration and a minor in film studies. My current thesis is on learning and teaching through visualization. I also made some short films over time, and I wrote a sci-fi murder mystery set in 2050.

Spotlight: one from each

003 selected

OSlings

2026

A Rustlings-style CLI that teaches operating systems by making you build a RISC-V kernel, one exercise at a time. The test suite boots your kernel in QEMU and reads the serial console, so a green test means it actually ran.

RustRISC-VQEMUno_std

Valentine

12 min

Writer, Director, Cinematographer, Editor, Actor. Screened at Artists’ Television Access in San Francisco.

Narrative shortCrew of 7

The 12:32 Murders

298 pp

Seven bodies in a month, every heart stopping at 12:32 AM, in a city that sells immortality by the vial. A murder mystery about what technology does to justice.

NovelNotion Press

Most of what I build is a way in

My current thesis is on teaching and learning through visualization. Since 2024 I have been a teaching assistant for five courses at the University of San Francisco, from beginner and intermediate Java all the way through computer architecture, operating systems and natural language processing. I spent a year as a tutor at the CS tutoring center, and a semester with Breakthrough SF at the San Francisco Day School as a volunteer, teaching coding to middle schoolers in the Python programming language.

That shows up in what I make. OSlings came out as a tool for going over concepts from operating systems, and it teaches the subject by handing you an empty kernel. Muxwriter came out of writing screenplays, and the desire of having a brainstorm companion to hear you out on your planning and narration. The sparse autoencoder came from wanting to see what a language model is doing rather than trusting its output. Focusix7 came from the idea of rewarding individual efforts for grinding, "locking in" and getting work done. And I strive to keep working on more such projects, creating pathways for more interesting and productive experiences.

Where I am

I am at the University of San Francisco, finishing a BS in computer science with an AI concentration and a film studies minor in 2027, and an MS in computer science in 2028. Dean's Honor Roll, six semesters.

Before this I built an AI conversation training agent at ClearOne Advantage, where employees rehearse live phone calls against a simulated client over real telephony.

I am looking for grad-level internships and new-grad roles for summer 2027.