Mustang

🎯 AI enthusiast · LLM apps · RAG · Full-stack (Python / React)

AI / ML & LLMs
- Python · basic PyTorch/TensorFlow
- Meta Llama 2 integration
- Hugging Face ecosystem (transformers, pipelines)
- Prompt design & basic evaluation
Web & Backend
- React · TypeScript
- Flask / FastAPI · REST APIs
- PHP · MySQL · MongoDB
Dev & Workflow
- Git · GitHub
- GitHub Actions (CI)
- Docker (learning)
- Jupyter, VS Code

Adaptive Learning Platform
Full-stack AI education platform built with React/TypeScript (frontend), Python (FastAPI/Flask) + MongoDB (backend), and Meta Llama 2 via Hugging Face for question-answering and personalised learning recommendations. Shows experience with LLM integration, REST APIs, and data-driven learning features.
LLM Chatbot Testing Tool (Ongoing Project)
Ongoing project for a Python toolkit that runs prompts from CSV/Excel through multiple LLM chatbots (e.g. HKChat + Perplexity), captures outputs, and will later use an LLM-based evaluator to automatically mark each test as PASS/FAIL based on expectations.
AI Content Marketplace Prototype
Small Flask-based prototype of an AI-assisted content marketplace, with endpoints to upload content and (planned/initial) AI-powered summarisation and tagging. Focused on building simple, clean REST APIs and experimenting with AI for content workflows.
Communication and Assignment Sharing Platform
Team project: full-stack web app for student chat, assignment sharing, and file uploads, built with React/TypeScript frontend and PHP + MySQL backend. Includes JWT-based authentication, REST APIs, and a structured database schema for users, messages and assignments.