Hi, my name is
Tharun Kumar.
I build cool software.
I am a problem solver specializing in building cool full-stack applications and solving interesting problems using AI. Currently, I'm pursuing a Bachelors in Computer Science at Princeton University.
About Me
My name is Tharun Kumar, and I enjoy analyzing data across the internet to make viable predictions that have a meaningful impact on society. My interest in Computer Science started when I was ten years old, when I built modifications to Minecraft to add new blocks and items. Although clueless at the time, I've grown significantly in the field of Computer Science through college-level courses and real-world experience.
Fast-forward to today, and I’ve had the privilege of building startups with Sebastian Thrun, working at OpenAI, and a music start-up. My main focus these days is building AI tools that transform the world while completing my Bachelors at Princeton University.
Here are a few technologies I’ve been working with recently:
- Typescript
- React, Next.js
- PyTorch, Tensorflow
- Python
- AWS
Some of my Experiences
Founder @ Sebastian Thrun Incubator
May 2024 - August 2024
Worked under direct mentorship from Sebastian Thrun with grant-based funding.
- Built viral SwiftUI application for deep-faking friends via voice-cloning, lip-syncing, and movement. Used 3D-reconstruction and pose estimation/mapping, fine-tuned lip-syncing models to achieve viral UGC. Available on the iOS App Store (trystorm.app).
- Developed novel video-splicing, editing, and collaging algorithm, and integrated into a full-stack SwiftUI application delivered on the iOS App Store (memorie.app). Partnered with large TikTok and Instagram influencers for go-to-market strategy.
- Built LoRA fine-tuning library hosted on scalable H100 infrastructure, and full-stack iOS application to generate personalized, hyper-realistic images of users for social media apps. DateMax AI is pending release the iOS App Store (datemaxai-web.vercel.app).
Some Things I’ve Built
Featured Project
Brainfeed - a productive feed
Brainfeed is a iOS app with a hyper-personalized and summarized newsfeed based on current events. Your feed is generated based on a SOTA multi-armed bandit recommendation system. Scrapes and generates ~500 posts per day. Built using SwiftUI, Typescript/NodeJS, Google Cloud, Firestore, and Pinecone. Try it out on Testflight!.
- SwiftUI
- Typescript/NodeJS
- Firestore/GCP
- Pinecone
Featured Project
Texas Climate Analysis
During the winter of '21, Texas had an unexpected winter surge that shattered nearly all power infrastructure throughout Texas. This catastrophic event prompted me to find a more promising solution to Texas's aging power grid. I constructed a Machine Learning model that analyzed climate data across Texas to construct a regionalized renewable power-grid. If implemented, the project would decrease Texas carbon emmissions by nearly 40%.
- Python
- Requests
- Plotly
- Pandas
Featured Project
Mood Detection
A prototype mood-analysis algorithm that detects human emotion through a basic conversation between the computer and person. Analyzed large amounts of data from a variety of Reddit subreddits to develop a Neural network capable of analyzing any conversation text from the user.
- VS Code
- Reddit API
- Python
- NLTK
- Keras
Noteworthy Research
view the archiveImplementing Django and React
Learned how to utilize Django to construct back-end servers to handle data and utilize Python Machine Learning algorithms. Implemented a To-Do list with a combination of a Django RESTful API and front-end React.
3D-Modeling in Python
Learning about how to 3D-model in Python based on basic object information.
Custom Object Detection
Utilized Tensorflow's immense Object Detection library to train a Custom Object Detection model that differentiated between animal masks. Learned how to obtain and clean data for Object Detection, as well as how lxml files are used in Tensorflow.
Facial Detection in Drones
Investigated how Artificial Intelligence could be paired with drones for Computer vision, facial tracking, and object detection. Programmed a drone to follow a person's face.
Learning Git
Researching how Git can be used in my personal projects and in the professional world. Learned and applied the skills in my current portfolio.
Adding Personality to Chatbots
Researching how Artificial Intelligence can be used to extract and embed personality in ordinary chat-bots. Discusses an ongoing project dedicated to the cause.
What’s Next?
Get In Touch
Whether you have a question or just want to say hi, I’ll try my best to get back to you!
Say Hello