Selected Projects
Pounds — AI Business Platform for Service Pros
Pounds is an all-in-one AI-powered business platform built for hands-on service professionals — nail technicians, barbers, trainers, and therapists. The platform unifies bookings, Stripe payments, a loyalty rewards programme, referral automation, and AI-generated social media marketing into a single dashboard. The embedded AI continuously learns from campaign performance, optimising content, timing, and outreach so practitioners can focus on their craft rather than client acquisition.
Yoh — AI Video Studio for Creators
Yoh is a direct-to-fan AI video studio that empowers creators to produce, publish, and monetize episodic content without a production crew. A single prompt is transformed by the AI story engine into full storylines with plot structure and character arcs. The platform handles video generation, character consistency across episodes, dialogue, original scores, and ambient audio — then delivers everything through a creator storefront where fans subscribe and stream. Creators keep 85% of every sale with weekly Stripe payouts.
Subscription Landing Page
This project is a high-performance landing page tailored for subscription-based services, designed to captivate potential customers and boost conversion rates. Built with Next.js and React, it features a responsive, intuitive interface paired with clear CTAs (calls-to-action) and a seamless user experience. Whether for SaaS platforms, content creators, or online courses, this landing page adapts effortlessly, leveraging modern design and technology to showcase product value and drive subscription growth.
Point trading system - staff end
This project is a point redemption system tailored for offline events, specifically designed to meet the operational needs of staff at merchant event venues. It features an efficient staff-facing interface, enabling personnel to quickly verify participants’ points and complete the redemption process. The application’s technical focus ensures stable performance even in high-traffic event environments. From scanning codes to instantly updating point records, this tool enhances on-site management efficiency and accuracy, ensuring smoother event operations.
Point trading system - user end
This project is a user-facing point application designed for offline events, allowing participants to earn points through mobile-based tasks or games. Points can be redeemed on-site for products or rewards in real time. The app features an intuitive and seamless interface, ensuring users can quickly view tasks, check point balances, and complete redemptions during the event. This system enhances participant interaction and engagement, making events more captivating and enjoyable.
Purchase ticket
This project is a convenient ticketing system that enables users to purchase tickets for events they’re interested in and manage their ticketing information directly within the app. We’ve developed a simple and user-friendly interface that supports ticket selection, payment, and the storage and viewing of e-tickets. The system ensures fast loading and secure transactions, allowing users to stay on top of their ticket status anytime, anywhere. From purchasing tickets to attending events, this app provides a one-stop solution, enhancing the convenience of event participation.
Swift Multilingual Voice Assistant Project
Developed a real-time voice assistant application powered by Next.js and AI technologies. This project integrates cutting-edge artificial intelligence, combining speech recognition, natural language processing, and speech synthesis to deliver a bilingual (English/Traditional Chinese) interactive experience.
- Integrated Groq API for efficient speech-to-text conversion and Llama3 large language model processing
- Leveraged Cartesia's Sonic TTS technology for smooth speech synthesis output
- Implemented real-time Voice Activity Detection (VAD) to optimize user voice interaction
- Built an extensible multilingual support framework with English and Traditional Chinese language switching
- Engineered error handling and fallback solutions for third-party API limitations to ensure stable user experience
- Used React Hooks and Web Audio API to implement efficient audio buffering and playback
Real-time Speech Transcription and Voice Command Control System
A web application that combines real-time speech transcription with intelligent voice commands. By integrating Deepgram's speech recognition technology, the system can convert speech to text in real-time and support speaker diarization for multi-person conversations. Users can activate voice command mode using the wake word 'Hey Summer', enabling hands-free operation. The system supports bilingual commands in both English and Chinese, coupled with dynamic visual feedback for an intuitive user interface.