HN user

Tetraslam

49 karma

meet.hn/city/42.3554334,-71.060511/Boston

Socials: - github.com/tetraslam - linkedin.com/in/shreshtbhowmick - x.com/tetraslam - https://tetraslam.world

---

Posts4
Comments26
View on HN

I'm an ML researcher @ MIT and a student at Northeastern University majoring in CS and Linguistics. I've built large-scale, efficient software for my neuromechanics and quantum photonics labs, for satellites, and for agentic video editing startups. I'm looking for full-time roles in language model training, agentic systems, full-stack (I have relatively good product sense), inference, and training VLAs for robots. My most recent projects: I won HackMIT with an autonomous power grid with iron-air batteries and mixed-integer linear programming-based electrical routing. Also built https://github.com/tetraslam/SHFLA; it hit the front page of HN. Over the summer I interned at https://mosaic.so, where I built their developer API, 4 core tiles (intro, outro, watermark, and AI augment), universal media search (search over hundreds of thousands of videos/images/audio in <2s accurately), and social media agents which autonomously find, edit, and post content to great amounts of engagement (>10k impressions per post in testing)

Location: San Francisco, CA

Remote: No

Willing to relocate: No

Technologies: Python, Zig, C, C++, Langchain/Langgraph, Javascript, Linux, Supabase, Redis, Pytorch, Slurm, CUDA, HPC, FastAPI, Embedded Linux, Tinygrad, Concurrency, Triton, Bash, Typescript, NextJS, LeRobot, Prime-RL, Modal

Resume: https://tetraslam.world/resume.pdf

Website: https://tetraslam.world

Email: shresht@mit.edu

I'm an ML researcher @ MIT and a student at Northeastern University majoring in CS and Linguistics. I've built large-scale, efficient software for my neuromechanics and quantum photonics labs, for satellites, and for agentic video editing startups. I'm looking for co-ops (8 months full-time, Jan - August) in systems engineering, ML engineering and research, agentic systems, full-stack, GPU infra and programming, and embedded systems. My most recent projects: I won HackMIT with an autonomous power grid with iron-air batteries and mixed-integer linear programming-based electrical routing. Also built https://github.com/tetraslam/SHFLA; it hit the front page of HN. Over the summer I interned at https://mosaic.so, where I built their developer API, 4 core tiles (intro, outro, watermark, and AI augment), universal media search (search over hundreds of thousands of videos/images/audio in <2s accurately), and social media agents which autonomously find, edit, and post content to great amounts of engagement (>10k impressions per post in testing)

Location: San Francisco, CA / Boston, MA

Remote: No

Willing to relocate: Within the United States, Singapore, and Japan.

Technologies: Python, Zig, C, C++, Langchain/Langgraph, Javascript, Linux, Supabase, Redis, Pytorch, Slurm, CUDA, HPC, FastAPI, Embedded Linux, Tinygrad, Concurrency, Triton, Bash, Typescript, NextJS, TailwindCSS, Langchain

Resume: https://tetraslam.world/resume.pdf

Website: https://tetraslam.world

Email: shresht@mit.edu

I'm an ML researcher @ MIT and a student at Northeastern University majoring in CS and Linguistics. I've built large-scale, efficient software for my neuromechanics and quantum photonics labs, for satellites, and for agentic video editing startups. I'm looking for co-ops (8 months full-time, Jan - August) in systems engineering, ML engineering and research, agentic systems, full-stack, GPU infra and programming, and embedded systems. My most recent projects: I built https://github.com/tetraslam/SHFLA and won the MIT Media Lab hackathon. It hit the front page of HN, and got me demo requests from Stephen Wolfram, David Sinclair, Joscha Bach, and Suno AI. I also built https://mosaic.so's developer API, 4 core tiles (intro, outro, watermark, and AI augment), universal media search (search over hundreds of thousands of videos/images/audio in <2s accurately), and social media agents which autonomously find, edit, and post content to great amounts of engagement (>10k impressions per post in testing)

Location: San Francisco, CA / Boston, MA

Remote: No

Willing to relocate: Within the United States, Singapore, and Japan.

Technologies: Python, Zig, C, C++, Langchain/Langgraph, Javascript, Linux, Supabase, Redis, Pytorch, Slurm, CUDA, HPC, FastAPI, Embedded Linux, Tinygrad, Concurrency, Triton, Bash, Typescript, NextJS, TailwindCSS, Langchain

Resume: https://tetraslam.world/resume.pdf

Website: https://tetraslam.world

Email: shresht@mit.edu

I’m an ML researcher/engineer at MIT + CS/Linguistics undergrad at Northeastern, looking for freelance roles or internships/co-ops (Jan–Aug 2026) in ML, backend, full-stack, or systems. Also willing to be recruited on an O-1 to the right startup.

Currently interning at a YC W25 startup Mosaic (usemosaic.ai) where I:

- Built the public API, our MCP server, and a video agent that turns longform content into shorts with auto captions and posts them

- Helped scale us to 100+ paying users in closed beta, 4k on the waitlist, and we're launching before my internship ends so you'll probably hear about us soon

Also built SHFLA (https://github.com/tetraslam/SHFLA), a fractal music language that won the MIT Media Lab hackathon, hit HN front page, and got me demo requests from Stephen Wolfram, David Sinclair, Joscha Bach, and Suno AI.

I move fast, write clean code, and deliver 10x what you pay me. I'm also very fun to hang around with and work with! Have several references who can vouch :)

Location: Boston, MA (in San Francisco for the summer)

Remote: Yes, or hybrid in Boston, but willing to get an O-1 and move to SF/NYC for the right startup/company (already fulfill 5 criteria)

Willing to relocate: Anywhere in the US

Technologies: Python, Zig, C/C++, PyTorch, CUDA, Supabase, Redis, FastAPI, NextJS, Slurm, Linux, Tinygrad, Typescript, TailwindCSS, Langchain, HPC, Vite, uv

Résumé/CV: https://tetraslam.world/resume.pdf

Email: shresht@mit.edu

hey everyone! i built https://onemonth.dev to streamline the process of searching for and vetting learning resources for my dad and some of my friends who aren't really great at searching for things on their own. i've found that i actually get results that make me want to use it myself, and so i thought i'd post the first version i have on here for feedback!

here are some of the features i'm currently building out and planning to release in <2 weeks: - pdf exports - shareable curriculum links - chat with/edit curricula - progress journal

looking forward to hearing what y'all think!

I'm an ML researcher @ MIT and a student at Northeastern University majoring in CS and Linguistics. I've built large-scale, efficient software for my neuromechanics and quantum photonics labs, for satellites, and for cloud computing startups. I'm looking for internships (4 months full-time, May - August) in systems engineering, ML, full-stack, and backend engineering. My most recent project: I built https://github.com/tetraslam/SHFLA and won the MIT Media Lab hackathon. It hit the front page of HN, and got me demo requests from Stephen Wolfram, David Sinclair, Joscha Bach, and Suno AI.

Location: San Francisco, CA

Remote: Yes, or onsite

Willing to relocate: Within the United States

Technologies: Python, Zig, C, C++, Javascript, Linux, Supabase, Redis, Pytorch, Slurm, CUDA, HPC, FastAPI, Embedded Linux, Tinygrad, Concurrency, Triton, Bash, Typescript, NextJS, TailwindCSS, Langchain

Resume: https://tetraslam.world/resume.pdf

Website: https://tetraslam.world

Email: shresht@mit.edu

Applied to ~45 jobs directly and got 2 interviews (my resume's fine, i think). posted here last month and got 8 emails within a week -> 3 offers. All at great companies/startups, uber-competent teams, and high-paying roles (like, double what public info shows). complete surprise too, since I was only looking for internships! Definitely a sample of 1, but I'm honestly completely flabbergasted by how helpful that one post was.

My Time at MIT 1 year ago

Huh. I suppose I should see if I can join Manolis Kellis' lab; I got invited to one of the salons he hosts at his house on the quantum basis of reality and it was really fun! He also demoed a tool his lab is building for genomics knowledge graphing, and I was struck by how incredibly well-designed and empirical the whole project was.

I'm an ML researcher @ MIT and a freshman at Northeastern University majoring in CS and Linguistics. I've built large-scale, efficient software for my neuromechanics and quantum photonics labs, for satellites, and for cloud computing startups. I'm looking for internships (4 months full-time, May - August) in systems engineering, ML, full-stack, and backend engineering. My most recent project: I built https://github.com/tetraslam/SHFLA and won the MIT Media Lab hackathon + Harvard C House hackathon. It hit the front page of HN, and got me demo requests from Stephen Wolfram, David Sinclair, Joscha Bach, and Suno AI.

Location: Boston, MA / San Francisco, CA

Remote: Yes, or onsite

Willing to relocate: Within the United States

Technologies: Python, Zig, C, C++, Javascript, Linux, Supabase, Redis, Pytorch, Slurm, CUDA, HPC, FastAPI, Embedded Linux, Tinygrad, Concurrency, Triton, Bash, Typescript, NextJS, TailwindCSS

Resume: https://tetraslam.world/resume.pdf

Website: https://tetraslam.world

Email: shresht@mit.edu

I'm a student at Northeastern University majoring in CS and Linguistics. I've built large-scale, efficient software for my neuromechanics and quantum photonics labs, for satellites, and for cloud computing startups. I'm looking for internships (4 months full-time, May - August) in systems engineering, ML, and backend engineering. My most recent project: I built https://github.com/tetraslam/SHFLA and won the MIT Media Lab hackathon + Harvard C House hackathon. It hit the front page of HN, and got me demo requests from Stephen Wolfram, David Sinclair, Joscha Bach, and Suno AI.

Location: Boston, MA

Remote: Yes, or onsite

Willing to relocate: Within the United States

Technologies: Python, Zig, C, C++, Javascript, Linux, Supabase, Redis, Pytorch, Slurm, CUDA, HPC, FastAPI, Embedded Linux, Tinygrad, Concurrency, Triton, Bash

Resume: https://tetraslam.world/resume.pdf

Website: https://tetraslam.world

Email: bhowmickshresht@gmail.com

I'm a student at Northeastern University majoring in CS and Linguistics. I've built large-scale, efficient software for my quantum photonics lab, for satellites, and for cloud computing startups. I'm looking for internships or co-ops in systems engineering, ML, and backend engineering.

My most recent project: I built https://github.com/tetraslam/SHFLA and won the MIT Media Lab hackathon + Harvard C House hackathon. It hit the front page of HN, and got me demo requests from Stephen Wolfram, Joscha Bach, and Suno AI.

Location: Boston, MA Remote: Yes Willing to relocate: Within the United States Technologies: Python, C, C++, Nim, Javascript, Linux, AWS, GCP, Azure, MongoDB, Redis, Pytorch, Slurm, CUDA, HPC, FastAPI, Embedded Linux, Tinygrad, Concurrency, Triton, Jester Resume: https://tetraslam.world/resume.pdf Email: bhowmickshresht@gmail.com

I know what you mean, and I'll work on it :)

SHFLA is Turing-complete though, so I am hoping people see that it has more potential than just being a music visualizer as I keep improving it! I'm currently in the process of rewriting it in Nim using SDL2, so once I get it to a performant state I'm going to implement an information-as-music encoder and all that.

Hi! I'm a college freshman at Northeastern, and this is my first hackathon project for a major hackathon (MIT Media Lab). It won the Unconventional Computing track prize, so I'm pretty happy! I made this because I'm a major conlanger, and I was wondering if it would be possible to think in terms of sound when looking at an image. Fractals have easy-to-map parameters, so I created SHFLA, a language which takes in music, and creates fractals based on 0.1 second (you can change this) chunks of music! It's Turing-complete, so you can technically encode a ridiculous amount of information and computation in my system, although writing it out as music might take a while.

Hope you find the project cool :)