I love working on projects that have pragmatic value. I’ve been thoroughly involved in founding as well sustaining startups that target fundamental problems like optimal resource allocation, smart waste management. Check them out!
BinIt: An AI-OS for Waste Recovery
I cofounded BinIt - the most accurate computer vision based waste classification software solution known to us. Ideal for waste recovery facilities to do accurate bale pricing, anomaly detection, tracking and quality standardisation. BinIt has won the Columbia Design Challenge, the Columbia Venture Competition and the Ignition Grant! We are also a proud alumnus of the Almaworks accelerator.
TokenTest: A pen-testing tool for proof-of-work blockchains
A winner of the IBM Blockchain Cyberdefense challenge - TokenTest will take the parameters of a proof-of-work blockchain protocol, and produce insight on potential and emminent mining attacks, projecting potential losses.
LionBase: a student-run Data Science Consultancy
I was part of the exec team of Lionbase a student-run group that provided data insight and solutions to small and midsize startups. Lionbase won the Columbia Venture Competition. Unfortunately due to the pandemic, we had to shut down the business.
Covid Sentiment Analysis
Worked on under the guidance of Dr. Juan Banda, we used natural language processing algorithms to perform sentiment analysis of public opinion of political leaders during the conception of the pandemic. Part of the codebase is available here
Convergence rates of Dimensionality Reduction Algorithms
As part of a project for COMS 4772 Unsupervised Learning, we conducted an empirical exploration of the convergence rates of tSNE and UMAP for benchmark datasets. Read the report here and code
Linear Regression Analysis to predict factors affecting movie quality
As part of a Linear Regression class, we performed a detailed analysis of movie ratings and revenue data to determine potential action areas for movie studios to focus on. Here’s our rather hurriedly made presentation :)
Portfolio Optimisation Basics
Made with Columbia’s Data Product Initiative, we created a tool that produced the latest correlation data for chosen tickers. Code here