Embedded Files
Snake (MIPS assembly, MARS)
TLDR: Snake Game implementation from start to finish coded using MIPS assembly language
Built a snake game entirely in MIPS assembly on the MARS simulator. To raise the challenge, a black wall appears after each apple (capped at 10) and apples despawn/respawn after ~2–3s via a busy-wait timer (timing varies with CPU/OS). I spent most effort on edge cases: never place walls in front/next to the snake, never spawn apples on walls or at the exact edges. The project reinforced low-level thinking around memory layout, grid math, collision detection, RNG bounding, and timer logic.
Sudoku Solver (C++ Backend and Python Frontend)
TLDR: This is a program that solves Sudoku puzzles automatically using a C++ backend for speed, and a Python frontend for the GUI.
Implemented a hybrid 9×9 Sudoku solver that first runs a deterministic deduction pass (fills any cell with a single certain value it knows goes there) and only then falls back to depth-first backtracking. This reduces the branching factor compared to naive backtracking. Leveraged AI coding assistants to accelerate a GUI implementation in Python to run the C++ code.
The Audio Hub (Full-stack web app)
TLDR: Full-stack web app with REST API + relational data model using FastAPI, SQLAlchemy, PostgreSQL, and React.
With a team, built and deployed a full-stack web app on Fly.io using Docker.
Backend: FastAPI + PostgreSQL (on Supabase) with typed schemas and modular routes.
Auth: OAuth2 with JWT (bearer); passwords were hashed before storing them on the database using bcrypt, while tokens were verified per request.
Data model: normalized tables with foreign keys linking each audio file to only one user.
Frontend: Next.js + TypeScript UI.
theaudiohub.fly.dev (might need a cold start)
Stock Prediction Web App (Server-trained models)
TLDR: A full-stack web app that trains ML models on the server and showcases them on the dashboard.
Built a full-stack app where prediction models are trained on the server, stored/queried via PostgreSQL-backed APIs, and presented in a React UI. Integrated two approaches (logistic regression and an LSTM-based simulation) and exposed prediction outputs through REST endpoints for end-to-end usage.
Tech: FastAPI, SQLAlchemy, PostgreSQL, React, Logistic Regression, LSTM-based simulation
Page updated
Google Sites
Report abuse