04 · Projects

Things I've built.

2026

1 project
Full-Stack · Academic Project · May 2026

Next-Step: Full-Stack School Management Platform

Full-stack school management platform built as a 5-person academic project. Async REST API with FastAPI exposing 20+ endpoints across 10 router modules backed by 9 SQLAlchemy ORM models (cascade deletes, JSON columns, unique constraints), with Alembic migrations on PostgreSQL and auto-generated Swagger / ReDoc docs. JWT auth with Argon2 hashing and Google OAuth 2.0 PKCE flow drives role-aware access for Admin, Teacher, and Student via factory-pattern middleware (get_current_user, require_admin, require_roles). Cross-platform React Native + Expo app (iOS, Android, Web) with TypeScript, Zustand state management, file-based routing via Expo Router, and AsyncStorage session persistence. Google Drive API integration for document management, Celery + Redis with Flower for background task monitoring. Co-developed via feature-branch Git workflow with 11+ pull requests, pytest + httpx async integration tests, and AI-assisted development tooling.

FastAPIPostgreSQLSQLAlchemyAlembicCeleryRedisFlowerJWT

2024

2 projects
AI/ML Core · Game AI & RL · Academic Project · November 2024

TrackMania Reinforcement Learning Agent

Autonomous racing agent in PyTorch implementing Implicit Quantile Networks (IQN) with NumPy-based state processing, reaching 85% track completion through iterative reward shaping. Training loop optimized with CUDA-accelerated parallel environment simulations, cutting training time by 60%.

PythonPyTorchIQNReinforcement LearningDeep RLCUDANumPydxcamGitHub →
AI/ML Core · Creative AI & Image Processing

Interactive Image Mosaic Generator

Built a Gradio app that generates artistic image mosaics using vectorized NumPy pipelines, giving users a fast, interactive UI to control mosaic density and style.

PythonNumPyGradioImage ProcessingComputer VisionVectorized OperationsGitHub →Live demo →

2023

1 project
Data Science · Social Media Analytics

Twitter Sentiment Analysis (NLP Project)

Built an end-to-end sentiment analysis pipeline over Twitter data, combining API ingestion, cleaning, feature extraction, and classical ML classifiers for topic-level sentiment.

PythonTwitter APINLPMachine LearningPandasNLTKTextBlobGoogle ColabGitHub →
Reach out

Got a question? Let's talk.