Skip to content
View raman2221's full-sized avatar
1 follower · 4 following
  • 04:25 (UTC -12:00)

Block or report raman2221

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Content in all repositories owned by your account will be closed.
Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
raman2221/README.md 
██████╗  █████╗ ███╗   ███╗ █████╗ ███╗   ██╗     ██╗███████╗███████╗████████╗
██╔══██╗██╔══██╗████╗ ████║██╔══██╗████╗  ██║     ██║██╔════╝██╔════╝╚══██╔══╝
██████╔╝███████║██╔████╔██║███████║██╔██╗ ██║     ██║█████╗  █████╗     ██║
██╔══██╗██╔══██║██║╚██╔╝██║██╔══██║██║╚██╗██║██   ██║██╔══╝  ██╔══╝     ██║
██║  ██║██║  ██║██║ ╚═╝ ██║██║  ██║██║ ╚████║╚█████╔╝███████╗███████╗   ██║
╚═╝  ╚═╝╚═╝  ╚═╝╚═╝     ╚═╝╚═╝  ╚═╝╚═╝  ╚═══╝ ╚════╝ ╚══════╝╚══════╝   ╚═╝

Ramanjeet Singh: CS at UMich, machine learning and edge AI

I'm Ramanjeet Singh, a computer science student at the University of Michigan. I train computer-vision models and get them running on real hardware. Most recently, I fine-tuned a bat detector on 10,000+ images and deployed it to an Arduino UNO Q at MHacks 2026.

LinkedIn: ramanjeets22 Email: ramanj@umich.edu 

╭──────────────────────────────────────────────────────────╮
│  now      → edge ai: getting models small enough to ship │
│  latest   → smart-bat-box (mhacks 2026)                  │
│  learning → object detection, quantization, onnx         │
╰──────────────────────────────────────────────────────────╯

things i've built

  • smart-bat-box: a bat box camera that tells bats from people. I fine-tuned BatNet (Faster R-CNN) in PyTorch: bat detection 3% → 70%, people ignored 96% of the time, plus a tiny model running on an Arduino UNO Q in ~10 ms/frame.
  • micrograd-lab: Karpathy's micrograd autograd engine, extended with Adam and SGD with momentum, new activations and losses, iterative backprop for deep graphs, tests, and CI.
  • ng-video-lecture: a GPT built from scratch in PyTorch, following Karpathy's Neural Networks: Zero to Hero lecture.

Tech stack: Python, PyTorch, TensorFlow, OpenCV, C++, JavaScript, React, Arduino, Git

training models, then putting them on real hardware



My GitHub contribution graph being eaten by a snake

Popular repositories Loading

  1. async-inference-engine async-inference-engine Public

    Distributed async inference pipeline with dynamic request batching (TypeScript, Python, C#, C++, RabbitMQ, Redis)

    Python 1

  2. chess-engine chess-engine Public

    C++

  3. real-sentiment-stock-tracker real-sentiment-stock-tracker Public

    Full-stack stock dashboard with news-sentiment signals, watchlist, and live sparklines (React + Express)

    JavaScript

  4. ramanjeet-singh-resume ramanjeet-singh-resume Public

    My personal resume website (HTML, CSS, JavaScript)

    HTML

  5. micrograd-lab micrograd-lab Public

    Forked from karpathy/micrograd

    Educational micrograd fork with Adam/SGD, training examples, texting-style profiles, and WhatsApp dataset preparation.

    Python

  6. ng-video-lecture ng-video-lecture Public

    Forked from karpathy/ng-video-lecture

    GPT built from scratch in PyTorch, following Karpathy's Neural Networks: Zero to Hero lecture

    Python