YouTube viewership forecasting
A model that predicted a video's reach from YouTube data, checked against what actually happened.
I'm a data science master's student at Columbia University. This summer I was a Data Science Intern at Artie, where I built models to forecast YouTube viewership and match brands with creators.
I spend a lot of my free time reading research papers and company engineering blogs to see how big teams solve problems. There's something new every day, and I like keeping up with it.
I'm doing my MS in Data Science at Columbia University. Before that I studied computer engineering at FCRIT in Navi Mumbai (CGPA 9.66/10) and published 5 papers in IEEE and Springer. Most of what I work on now is forecasting, causal inference and ML for real products.
I was a Data Science Intern at Artie this summer, and before that I interned at Deloitte and IIT Roorkee. I'm also a Graduate Student Assistant at the Northeast Big Data Innovation Hub at Columbia. Beyond the data, you'll find me wandering through Central Park or Riverside Park, hunting for new coffee shops and bookstores, and buying the most random things from Five Below. 🌸
Wrote and published papers on NLP, quantum computing and blockchain in IEEE Xplore and Springer.
Artie is a creator-economy analytics startup. I worked on two problems: predicting how many views a creator's video would get, and matching brands with the right creators.
A model that predicted a video's reach from YouTube data, checked against what actually happened.
A ranking engine that read a brand brief and found the creators whose content fit it best.
The languages, libraries and tools I use most.
Class projects, side projects, and a couple that turned into papers.
How extreme heat affects public health in the U.S. I cleaned the data and built charts in R, D3.js and ggplot2 showing heat-related hospitalizations and deaths by region and group.
View project
A mobile app that scans food labels with OCR, classifies the nutrition with a Random Forest, and suggests healthier swaps using cosine similarity. Published in Springer LNNS.
Read paper
Classifies ECG reports with a CNN (92% accuracy), then uses generative AI to explain the result in simple terms with lifestyle tips based on the patient's profile.
Movie and book recommendations using ALS collaborative filtering. It returns your top 10 picks.
A Bridgerton watch-party site for fun: a gen AI translator that turns your messages into Regency-era English, gossip polls, character quizzes and live chat. Hosted on Netlify.
Visit site
An AR app built in Unity and C#. You place 3D objects on real surfaces and watch how gravity and friction act on them.
A Django to-do app on AWS EC2 with auto-scaling. Multiple users can edit lists together in real time.
View code
Lets NGOs set up their own website with donations, event listings and volunteer sign-ups.
Five papers from my undergrad, published in IEEE Xplore and Springer.
Literature Survey
Looks at where quantum computing is being used today and what's holding it back.
View paper →Research Paper
Used LDA and NMF topic models on Instagram posts to find trends as they emerge.
View paper →Smart Contract Implementation & Research
Solidity smart contracts on Ethereum that manage grid load and payments between electric vehicles and the grid.
View paper →Technical Paper
The paper behind FoodMO, the OCR and ML app for checking food nutrients.
View paper →Literature Review
An overview of NLP methods, where they're used, and open problems.
View paper →A few things I've pushed recently.
Simulates how a patient might respond to treatment, using physiological data from several sources.
LLM agents that query an online retail dataset and pull out business insights.
Predicts house prices from property and location features with a neural network.
English to French translation with an LSTM encoder-decoder.
Predicts heart disease risk from clinical data using KNN.
Forecasts avocado prices from past sales and seasonal patterns using SVR.
I post on Medium about ML, causal inference and my projects, in plain language.
Read on MediumI'm looking for full-time data science and ML roles after I graduate in December 2026. If you're hiring, or just want to talk about forecasting or causal inference, email me.