Engineered a PyTorch neural network using Python to optimize pit stop strategy, leveraging Formula 1 telemetry data and accurately forecasting remaining tire life based on telemetry data. Designed and deployed an end-to-end ML pipeline including large scale data sourcing (using web crawling), data cleaning, data ingestion, preprocessing, custom feature engineering, circuit clustering (K-means), and hyperparameter optimization (Optuna).

Engineered a custom PyTorch neural network using Python for image classification, featuring parallel convolutional branches with expert-guided attention mechanisms and achieving 90%+ accuracy. Applied advanced ML techniques (MixUp, CutOut, LR scheduling) and developed a comprehensive data preprocessing/augmentation pipeline to enhance generalization and reduce overfitting.

Web-based Workout Tracking. Developed a custom workout tracking web app with a mobile-friendly interface using Next.js, React, TypeScript, and Tailwind CSS. Architected a PostgreSQL data warehouse using Prisma ORM to normalize and store workout sessions, sets, and personal records. Deployed the application on AWS, utilizing EC2, S3, RDS for scalable infrastructure and data persistence.

Developed a suite of Python-based CLI tools for quantitative financial analysis, including backtesting elementary trading strategies, risk assessments, and technical analysis. Engineered the analytical core using yfinance, pandas, and matplotlib to implement algorithms for key indicators (e.g., RSI, MACD) and risk metrics (e.g., VaR, Sharpe Ratio), complete with a candle-stick plotting feature.

Developed a web-based synthesizer application simulating a classic subtractive synthesizer using HTML, CSS, JavaScript, Tone.js, and AudioKeys. Implemented real-time audio synthesis with two oscillators, a modifiable filter, an LFO, and multiple effects (e.g., reverb, chorus, delay), integrating interactive controls to replicate a modern synthesizer in the browser.