Things I've Built
A mix of things I built because they needed to exist, capstones, hackathons, actual problems worth solving, and things I built just to see how something worked under the hood. Some are live, the rest are on GitHub for now.
An applicant tracking system I built as my capstone, but flipped: instead of giving employers tools to manage candidates, it gives candidates the same tools employers use to manage their hiring process. Candidates can apply, generate tailored resumes, and track their status through the pipeline, backed by AI that handles the busywork and offers feedback along the way.
Visit →A dashboard that watches for disease outbreaks, powered by three agents that hand work off to each other. A News Agent pulls from Google News RSS, a Health Agent pulls from CDC and CIDRAP RSS, and a Synthesizer Agent (Claude) combines what they find into structured signals shown as markers on the map.
Visit →A library management system to create curated reading wishlists and also manage book offerings. Admins can pull books in from an external API or add them manually, with duplicate checks before anything gets saved. Users build their own favorites list from the catalog with full filtering and search, and role-based access keeps admin tools separate from what regular users can do.
Visit →A handwritten digit recognizer built using only NumPy, no ML libraries. I wrote the neural net and backpropagation myself and got it to 97.84% accuracy on MNIST. There's a canvas where you can draw a digit and watch the neurons activate in real time.
View on GitHub →A visual simulator for a 5-stage CPU pipeline, built after struggling to understand pipelining in my computer architecture class. It shows instructions moving through the pipeline stages live, including how the CPU handles hazards and branch prediction.
View on GitHub →A Chrome extension that scores how sustainable a brand actually is while you shop. One click gives you a report card across environmental impact, labor practices, and transparency, powered by an LLM that also flags greenwashing when a brand's claims don't match its practices.
View on GitHub →A lexical analyzer, recursive-descent parser, and interpreter for a custom C-like language, built from scratch in C++. Tokenizes source code with a DFA, validates it against an EBNF grammar, then interprets and runs it in real time with symbol tables, type checking, and runtime error detection.
View on GitHub →