Selected work
A mix of machine-learning classifiers, finance tooling and lower-level systems work. Some are coursework taken further than it needed to go; some are just curiosity that got out of hand.
A model that predicts whether a stock closes up or down the next day from technical indicators (moving averages, RSI, volatility). A study in how far simple signals actually get you — and how easy it is to fool yourself.
An NLP classifier that flags phishing and spam from email text — TF-IDF features into a few different models to compare precision/recall. Grew out of my time with the USYD Cybersecurity Society.
A convolutional neural network trained on MNIST, with a small web canvas so you can draw a digit and watch it guess in real time. The classic — done properly, end to end.
A small HTTP server built from sockets up — thread pool, concurrent connections and careful handling of shared state. Came out of systems programming at USYD and the urge to know what's under the abstraction.
A dashboard that pulls price data and surfaces portfolio risk — volatility, drawdown and a simple valuation view. Where the finance degree and the code finally shake hands.
A small set of reusable pytest fixtures and helpers that cut the boilerplate out of writing tests — inspired by the kind of utilities I build day-to-day at Bayer.
More on the way — I'll keep this updated as I ship things.
Get in touch
Happy to walk through any of these — code, trade-offs and what I'd do differently.
pjimit716@gmail.com