Undergrad thesis. Exploring whether quantum circuits can meaningfully complement classical neural networks in real-world classification tasks. Designed and evaluated a parallel hybrid architecture combining a classical neural network with a small variational quantum circuit, tested across multiple medical datasets under rigorous cross-validation. Beyond standard accuracy comparisons, developed a diagnostic toolkit to probe what each branch of the model actually learns at the representation level, revealing that hybrid models can fail in fundamentally different ways depending on the dataset, and that no single architectural fix works universally. Used these insights to propose targeted, mode-specific variants that address each failure at the appropriate layer of the architecture.
Projects — 4
I touched these things.
Browse through my collection of tech projects, spanning from web apps to machine learning experiments. Each one is a snapshot of my learning journey and the fun (really?) I had building it.
Full-stack meal planning app: React frontend (weekly planner, paginated recipe cards, auto-generated shopping list, gamified streak tracking with 365-day activity heatmap) + Spring Boot REST API with JWT/BCrypt auth, Spring Data JPA, PostgreSQL, Flyway. Integrated Google Gemini AI for on-demand recipe generation and ingredient-based fridge search.
School project. Built responsive React UI for browsing and streaming pages in a team of 5 (SCRUM). Configured AWS S3 with presigned URLs for secure video upload/download without exposing credentials, plus bucket policies and CORS settings for FE↔BE communication.
School project. Implemented a POP3/SMTP email client over raw TCP sockets in Python with multi-threaded connection handling for concurrent send/receive. Local email storage with SQLite and configurable filtering rules for inbox management.