Companies Founded

Co-Founder

Campus Lens

campuslens.xyz ↗

Proprietary occupancy-tracking platform built for university clients. Computer vision system monitors real-time space utilization across campus facilities — live in 100+ locations at the University of Rochester.

100+
Locations
10+
Team Members
Uni Audit Passed
Co-Founder

Krundle

krunda.com ↗

AI voice agent platform that gives businesses across New York a customized voice. Developed financial models for pricing and forecasting, led a team of 10+, and saved clients a combined $70k+ through optimal staffing recommendations.

$70k+
Client Savings
10+
Team Members
NY
Market

Built & Shipped

Production

Newsliy

newsliy.org ↗

ML-powered news aggregation and recommendation platform processing 1,000+ daily articles from 500+ sources. Hybrid recommendation engine combining dense vector embeddings, BM25 ranking, and Thompson Sampling multi-armed bandit algorithm.

940
Articles/min
<5s
Cold Start
500+
Sources
Hackathon Winner

ClubSpot

Full-stack web platform enabling university clubs to share events and updates while fostering community engagement. Won DandyHacks — 1,500+ lines of code shipped within hackathon time constraints.

30+
API Endpoints
1.5k+
Lines of Code
Open Source

Hope For NYC

hopefornyc.com ↗

Free, open-source web platform mapping 4,000+ homeless service locations across NYC — shelters, food, medical care, warming centers, and more. Built to be privacy-first and offline-capable so both homeless individuals and the social workers who serve them can find help fast, with no barriers.

4,000+
Service Locations
Free
Always
Offline-Capable

Academic Work

Research Paper

Predicting Stock Price Movements Following Significant Market Changes

Jan 2025 – May 2025  ·  Simon Business School, University of Rochester

Led a cross-functional research team developing advanced ML models for post-volatility stock price forecasting. Architected a deep neural network achieving measurable predictive edge on S&P 500 stocks following significant market movements (>10%). Applied SMOTE for class imbalance, time series cross-validation, and permutation-based feature importance analysis.

TensorFlow Deep Learning Sentiment Analysis Financial Modeling SMOTE Time Series
58.07% Prediction Accuracy
135% Long-Only Returns
114% Long-Short Returns
22% Market Benchmark