I’m a Computer Science undergraduate from Mahindra École Centrale (MEC), India, currently teaching myself technology that interests me.

I regularly work with


MEC, India

Worked as a teaching assistant for the DataScience bootcamp by the CS department. Created interactive assignments on Jupyter Notebooks to teach various ML/AI concepts. Assisted students in debugging and improving their code. I also developed a submission portal and leaderboard to score day-to-day assignment submissions.

University Of Texas, Austin

Worked with a team of 4 to evaluate the effectiveness of pre-trained models and unsupervised learning techniques in detection of cancer. We started with simple unsupervised algorithms such as DB-Scan and later used MASK RCNN and its predecessors with pre-trained models to identify cancer cells.

MEC, India

Worked on a collision avoidance system for aircraft, using Genetic algorithms and stochastic optimization techniques in a simulated airspace while ensuring minimum deviation, smooth maneuvers and minimum fuel overhead.


Scrappy Engine

A game engine (under construction) built using C++ and Vulkan. It currently lacks the ‘game’ parts of a game engine, and the renderer is still under development. If you want a fancy triangle rendered at over 500FPS, this is a suitable ‘Game Engine’ for you. Project Link: Github


A URL shortener built using C# and Microsoft’s .NET Core platform. This is an ongoing project to accustom myself with .NET Core, networking, APIs, Unit tests and CI/CD systems. Project link: GitHub

Web Server

Built a secure, minimal docker container for nginx to host the previous version of my website. CI/CD builds are handled by Azure Pipelines. Project link: GitHub


ANND (ANN for Dummies) is a simple ANN library written using Python and Numpy. It has a simple API that allows users to create ANNs of arbitrary complexity. It’s aimed at making the concepts behind ANNs in code more readable and accessible. Project link: GitHub


Built a GAN (Generative Adversarial Neural network), to augment a dataset with unbalanced class labels. This network was trained on a proprietary industrial steel manufacturing dataset (with ~28% ‘breakout’ cases). The goal was to create a generator that can create similar data to that of the reference data set but with an even balance of labels (‘breakout’ and ‘no-breakout’). We obtained promising results, but further tests were necessary before it could be commercially deployed.


An Autonomous Maze solving robot built using a Raspberry Pi and a bunch of ultrasound sensors. Robot’s motion and sensing of surroundings are handled in parallel with the help of multiprocessing module. It helped improve the robot’s response time and made its movements less sporadic. The constraint was to use python, to demonstrate its broad applicability.


Worked on the server side of the College Chat Network. This was a coursework project where we decided to build everything ourselves to really expose everyone to the complexity behind something as mundane as a chat application. The message exchange server was deployed on GCP’s Ubuntu VM. Messages were serialized as JSON objects and exchanged using the WebSocket protocol with custom logic handling based on message headers. Client-Server data exchange is secured by HTTPS. We also partially implemented peer-to-peer chat in local networks but later dropped the idea as it didn’t scale well.


Built a hovercraft with styrofoam, ESCs and a hacked up controller. It competed with 15 other hovercrafts from various teams from around my college and bagged the first place. It was so fast that 3 other hovercrafts broke down trying to compete with this.

What I’m working on

  • Game engines and rendering
  • Scalable architecture; à la K8s and containers
  • Network architecture (DNS, nameservers etc.)
  • Information security and Cryptography
  • Embedded programming