PROJECT · 001
Local-first contact manager
- Developed a secure, local-first contact management application with a Python (FastAPI) backend and a vanilla JavaScript frontend.
- Implemented real-time search filtering, toggleable grid/list views, and responsive custom themes for cross-device use.
- Engineered a fully offline experience with locally stored database assets and automated browser launching on server start.
PythonFastAPIVanilla JS
PROJECT · 002
Native desktop study alarm system
- Built a local-first study task manager pairing a React/Tailwind CSS frontend with a Flask/SQLite backend to track daily lecture schedules.
- Used Python's
plyer and schedule libraries to trigger native desktop notifications and local audio alarms for time-sensitive tasks.
- Designed a responsive UI with Vite, letting users create, manage, and toggle study sessions in real time.
ReactTailwind CSSFlaskSQLiteVite
PROJECT · 003
RAG-powered search interface
- Engineered a Retrieval-Augmented Generation chat app combining Perplexity AI's real-time web search with Google Gemini's reasoning.
- Built a responsive web interface with Flask and Tailwind CSS showing search context, clickable source URLs, and summarized answers.
- Integrated robust API handling for seamless dual-model processing, with optional CLI support.
FlaskTailwind CSSGemini APIPerplexity API
DATASET · 004
Global Health Professions Student Survey
Kaggle ↗
Mar 28 – Mar 30, 2026
Public health data engineering
- Published a rigorously cleaned Kaggle dataset of 15,282 records on tobacco use and cessation training among medical students across 77 countries.
- Normalized data formatting by resolving floating-point artifacts, standardizing sample sizes, and fixing conflicting duplicate keys.
- Structured complex demographic and behavioral indicators for exploratory analysis and predictive modeling.
Data CleaningPandasKaggle
DATASET · 005
Mobile Addiction Behavioral Dataset
Kaggle ↗
May 27 – May 30, 2026
Predictive modeling resource
- Curated and published a dataset of 13,589 user records covering daily mobile usage indicators and self-reported behavioral metrics.
- Structured 11 integer-valued features — including screen time, app sessions, and notification volume — against a binary addiction target.
- Delivered a clean, missing-value-free data dictionary suited to classification, segmentation, and digital-stress analysis.
Data CurationBinary ClassificationKaggle