CV
Rohan Fozdar

Rohan Fozdar

Jr AI Engineer at Frontline Performance Group

About

I'm a junior AI engineer at Frontline Performance Group, where I build AI systems for hospitality brands including Marriott, Hilton, Jumeirah, and Hyatt. I studied pure mathematics and data science at Knox College, and I'm also founding a stealth startup.

I'm based in Chicago, and I'm interested in AI and machine learning architectures and how they get implemented. If you'd like to get in touch, use one of the links above.

Work

Junior AI Engineer

Frontline Performance Group · current

Building AI systems used by Marriott, Hilton, Jumeirah, Hyatt, and other hospitality brands.

Quantitative Analytics Intern

Monarch Networth Capital · Nov 2025 – Jan 2026

Designed a momentum trading strategy on NSE ETFs using time-trend derivatives. The backtest returned 45% with a 3.43 Sharpe on BankNifty over one year. A moving-average strategy, comparing the current market price with the moving average, returned 3100% with a 2.86 Sharpe on Nifty over 25 years.

Mathematics Teaching Assistant

Knox College Department of Mathematics · Mar 2024 – present

Wrote class material and led recitations for Calculus I–III, discrete math, linear algebra, foundations of analysis, methods of proof, differential equations, mathematical statistics, and financial math. Graded homework and exams, and hosted department tutoring sessions with the other TAs.

AI Intern

MeasureOne · Jun 2023 – Aug 2023

Fine-tuned early ChatGPT models on prompt–completion pairs so they could match client data tables and generate parser code. Built a preprocessing pipeline that normalized heterogeneous transcript formats and cut manual engineering workload by 25%.

Projects

Selected work. More lives on GitHub.

  1. 01

    ERCOT Electricity Forecaster

    XGBoost spot-price model trained on nine data sources from 2016–2024. Spike-rate accuracy of 58.8%, against 28.9% for the day-ahead market, with grid-stress signals up to 144 hours ahead. An extension using IBM Research's Tiny Time Mixers was in testing, set to conclude June 2026.

    code

  2. 02

    Promotion Gauger

    Sentiment tool for the Knox Consulting Club. It pulls Google News and Amazon reviews for a promotion and scores each source with a retail-tuned RoBERTa model on price perception, brand sentiment, and urgency.

    demo code

  3. 03

    AutoSearch

    Job-search platform that ingests listings, scores them against a résumé with FAISS and sentence-transformers, and ranks them with an ATS microservice.

    demo code

  4. 04

    Binomial Options Pricer

    Binomial pricer with live Yahoo Finance data, Treasury yields as the risk-free rate, and a 3D implied-volatility surface, built to show how implied volatility moves with the market. Increased enrollment in Financial Mathematics (Math 227) by 30%, and received Richter Grant funding from the Knox Department of Mathematics.

    website code