Projects
All projects



Big Data · NLP
YouTube Sentiment Analysis — Real-Time Big Data Pipeline
Real-time Big Data pipeline that streams YouTube comments through Kafka, classifies sentiment with a multilingual BERT model on Spark Structured Streaming, and indexes results in Elasticsearch — visualized on a Streamlit dashboard auto-refreshing every 30s. Processes 500+ comments per cycle with automatic deduplication, continuous 5-minute collection, and ~2–5 min end-to-end latency.




Full-Stack · Team Project
BookTrack ChurnGuard — Library Management Platform (Team of 3)
Team project (3 developers) — my role was Full-Stack Developer (Frontend/Backend). Built the visitor and user interfaces (Next.js 14, React 18, TypeScript), architected the FastAPI backend with Google OAuth + JWT authentication, modeled the PostgreSQL database, set up Docker for development, and deployed to production on AWS EC2 with Docker Compose. Shipped visitor, user, admin, and moderator dashboards — the moderator dashboard supervises the churn-prediction ML models (Scikit-learn, XGBoost) built by teammates, alongside automated retention campaigns and activity analytics — plus a documented RESTful API and a comments/recommendations system.


Full-Stack · Team Project · Oct. 2025 — Jan. 2026
Gestion de Stock — Inventory Management (Desktop App)
Team project (7 full-stack developers) — built the order management module: an interface for creating and tracking orders, the full order lifecycle (creation, validation, delivery), client- and server-side data validation, and role-based access (Admin, Warehouse Staff, Client).


