Data Scientist and Mathematician. I build things with data, code, and statistical rigour — from LSTM forecasting apps to contaminated chemical detectors.
I'm an MSc Data Science student at the University of Nottingham, with a First-Class Honours degree in Mathematics from the University of Delhi. I'm interested in the space where statistical theory meets real-world prediction problems.
My recent work spans from building a privacy-first file conversion tool (DeltaConvert) to training LSTM models for currency forecasting (Stochastix), to winning a statistical ML competition by proving that a two-parameter linear model can beat a Random Forest when the data structure is right.
I care about understanding why a model works, not just that it works.
Privacy-first file conversion tool — images, documents, OCR — all processed server-side. Built with Flask, deployed on Hetzner with Docker Compose, Nginx, and SSL via Certbot.
LSTM-based 30-day currency exchange rate forecaster. 29 trained models, React frontend with interactive Plotly charts, daily ECB audit cron job.
Chemicals: Per-type linear regression beating Random Forest (score 98, 1st of all groups). Trains: XGBoost on Sheffield–Nottingham increment (MSE 28,078). MATH4069 coursework.
PHP/MySQL hospital management system with role-based access, patient search, admissions, prescriptions, parking permits, and a custom audit trail. Graded Distinction at UoN.
Web scraping 2,000+ reviews with BeautifulSoup, NLP sentiment analysis, and Random Forest booking prediction model.
AI-powered financial chatbot parsing 10-K and 10-Q reports with rule-based logic to provide user-friendly corporate insights.
10-year exploratory analysis of 6 major banks' stock data (2014–2024) using Pandas, Matplotlib, and Seaborn.
Machine Learning, Statistical ML, Big Data, Databases, Time Series & Forecasting
South Asia Postgraduate Excellence AwardFirst-Class Honours — Probability, Statistics, Mathematical Finance, Linear Algebra
INSPIRE Scholarship — Top 1% nationallyQA on AI-generated mathematical models. Evaluated calculus and algebra outputs for hallucination and factual accuracy. Developed complex maths questions to improve model comprehension.
Built supervised and unsupervised ML models (linear regression, K-means clustering). Created detailed visualisations. Conducted 15 peer reviews and supported fellow interns.
Tutoring primary and secondary school students in mathematics. 100% pass rate in final exams. Focus on building problem-solving intuition through interactive, positive learning environments.