Click anywhere for a surprise
   

Georgia Tech · Computer Science

↗ resume ↗ linkedin ↗ github ↗ instagram

About Me

Hi! I'm Karshin Sachdeva and I am a student at Georgia Tech studying Computer Science, specialising in System Architecture and Intelligence.

I am particularly interested in machine learning and applying deep learning to real-world problems. From creating CNNs to help me cheat at GeoGuessr to building punch classification models, I love developing models that are both engaging and practical.

Outside of coding, I love watching Australian Football, hooping and playing guitar. I also particularly like listening to artists like Nine Vicious and Lisa Ono.

Karshin Sachdeva

Experience

Georgia Institute of Technology — Machine Learning Developer
Aug 2025 – Present · Atlanta, Georgia · Big Data Big Impact
Developed a (2+1)D CNN using PyTorch and NumPy to classify boxer punch types from video data and provide punching form analysis. Processed spatiotemporal data provided by MediaPipe and optimized performance using deep learning frameworks.
Georgia Institute of Technology — Software Team Member
Aug 2025 – Present · Atlanta, Georgia · RoboJackets
Improved inter-robot communication using C and enhanced robot dribbling performance with real-time control logic. Debugged ROS2-based software to allow for the use of RQT tools and robot controllers in virtual machine environments.
Learning Dimensions Network — Junior Tech Specialist
Aug 2023 – Dec 2023 · Melbourne, Australia
Digitized and streamlined scalable digital systems to transition in-person workshops to fully remote delivery. Developed adaptive solutions that enhanced virtual workshop performance and supported nationwide deployment.
Swinburne University of Technology — Research Leader
Aug 2022 – Dec 2023 · Melbourne, Australia
Designed and optimized experiments on microgreen fertility in zero gravity conditions after extensive data analysis. Developed an experiment using 25 vials sent to the International Space Station. Mentored a cross-disciplinary research initiative on mycelium-based infrastructure that was also sent to the ISS and featured in two documentaries.

Projects

Music for the Hearing-Impaired
↗ view
Mar 2021 – Present
Engineered a physical wristband prototype with Raspberry Pi that converts WAV files into haptic vibrations through a motor controller and stepper motors in Python. Developed an accompanying iOS app in Swift and Xcode, and a web app interface that engaged over 1,000 active participants from the Auslaners Deaf Community.
Senior Citizen Mobility App
↗ view
Mar 2024 – Dec 2024
Built a web app in HTML, CSS, and JavaScript that lets senior citizens personalize their mall shopping experience. Solved the Travelling Salesman Problem using A*, Dijkstra, and simulated annealing to minimize routing time complexity.
GeoGuessr Cheat
↗ view
Dec 2025 – Present
Built a CNN using ResNet that analyzes street-view image data to identify which country a GeoGuessr screenshot was taken in. Created a custom training dataset by scraping and labeling imagery via the Google Maps API.
Peacho - AI-powered Telecommuniactions Tool
↗ view
GTHacks Hackathon
Built a voice-based healthcare assistant where an LLM processes the conversation in real time to update records, answer provider-approved medical questions, and route booking requests based on availability. Developed the system using React and Firebase for the provider dashboard and EMR, with a Python backend connected to Twilio voice calls and AssemblyAI transcription.