I am a Computer Science student at The University of Hong Kong (HKU). My work focuses on bridging the gap between deep learning research and production-ready systems, with a particular emphasis on Medical AI, Computer Vision, and Machine Learning.

• 🔭 Current Focus: Engineering AI-powered diagnostic tools for healthcare.
• 🎓 Education: BEng in Computer Science @ HKU (Year 2).

An AI-powered stock analytics engine and full-stack trading dashboard implementing advanced statistical forecasting and data grounded financial AI advisor.

• Impact: Executes regime-conditional Monte Carlo simulations, GARCH volatility forecasting, and Hidden Markov Models, paired with an AI advisor generating real-time market insights with inline SEC EDGAR citations, News, and other market data.
• Stack: Python, FastAPI, Next.js 14, TypeScript, Supabase (PostgreSQL), hmmlearn, arch, Tailwind CSS.

A dual-architecture diagnostic platform utilizing EfficientNetV2-S for granular analysis and ConvNeXt-Tiny for high-stability classification.

• Impact: Capable of recognizing 44 distinct tumor signatures and 17 broad diagnostic groups with 98.0% accuracy, featuring a MedGemma 1.5 4B Vision-Language Model (VLM) for "Medical Peer-Review" and confirmed diagnosis.
• Stack: Python, PyTorch, TensorFlow, FastAPI, React, Docker.

A privacy-first, zero-database AI study assistant that converts local PDF documents into structured 4-zone study guides and interactive context-based chat sessions.

• Impact: Processes multiple PDFs fully within the browser to generate exportable Markdown cheatsheets (Deep-Dives, Analogies, Solved Problems, and Exam Traps) and provides isolated, step-by-step AI tutoring with zero data retention.
• Stack: Python, FastAPI, Gemma 4 31B IT, pypdf, Docker, Vanilla JS.

A real-time monitoring and multi-server alert pipeline tracking wildfires, earthquakes, and tropical cyclones across Thailand and Greater Indochina.

• Impact: Feeds parallel NASA and USGS telemetry streams into a low-latency alert pipeline, automatically broadcasting regional hazard events within 15 minutes to an auto-managed network of multi-tenant Discord channels.
• Stack: Next.js 16 (App Router), React 19, TypeScript, Tailwind CSS 4, Leaflet.js, Upstash Redis, Vercel.

An automated intelligence engine that leverages Gemini 3 Flash to synthesize critical developments in AI and Global News.

• Impact: Delivers curated briefings via a specialized Discord bot and a modern web interface.
• Stack: Python, Discord.py, Gemini API, RSS.

A tactical 1v1 strategy game engineered with a focus on advanced OOP principles and polymorphic class structures.

• Stack: C++, OOP, Makefile.

A premium, interactive technical profile featuring a physics-based 3D environment and fluid motion systems.

• Impact: Custom Three.js "wormhole mesh" style hero page implementation with interactive particle physics.
• Stack: React, Three.js, Framer Motion, Tailwind CSS.

Developed by Natthakitt Prapunwattana