AI Researcher / Full-Stack Engineer / Open-Source Builder
I'm a Computer Science undergraduate at The Hong Kong Polytechnic University, where I maintain a 3.85 GPA while shipping production-grade AI applications and publishing research at top-tier venues like KDD.
Selected for PolyU's prestigious Undergraduate Research and Innovation Scheme (URIS), my work spans multimodal AI, large language model fine-tuning, and building developer tools that turn cutting-edge research into real-world impact. I believe the best software is built where deep technical rigor meets genuine user empathy.
Never miss a breakthrough. Automatically tracks and curates the latest papers from top CS conferences and journals on arXiv, with intelligent filtering, categorization, and weekly digest delivery.
Your inbox, handled intelligently. A native Mac assistant that reads emails through Mail.app, generates concise summaries, and drafts context-aware replies — so you spend minutes, not hours, on email.
Journaling meets AI. A beautifully crafted mobile app that turns daily reflections into actionable insights — automatic mood tracking, intelligent scheduling suggestions, and personalized productivity patterns.
Data-driven investing, demystified. An ML-powered analysis engine that processes market signals, evaluates portfolio risk, and surfaces actionable investment opportunities with explainable AI reasoning.
Take the guesswork out of grad school. An intelligent recommendation system that matches students with programs, advisors, and funding opportunities based on research interests, academic profile, and career goals.
Citations on autopilot. A smart reference manager that auto-discovers relevant papers, suggests contextual citations as you write, and keeps your bibliography organized across multiple projects.
Constructed a comprehensive benchmark dataset enabling machines to read, synthesize, and generate academic survey papers. Addresses the growing challenge of information overload in scientific literature by automating the literature review pipeline end-to-end.
Investigating how vision-language models can be unified to achieve deeper cross-modal reasoning. Exploring novel architectures that bridge the gap between visual perception and linguistic understanding for real-world applications.
Designed and trained 2 custom LoRA adapters that specialize general-purpose LLMs for domain-specific tasks, achieving significant performance gains with minimal compute overhead — proving that targeted fine-tuning can outperform brute-force scaling.
Whether it's a research collaboration, an exciting project idea, or just a good conversation about AI and tech — I'd love to hear from you. Let's build something remarkable together.