Full-stack & AI Projects
PickNGo
Collaborated in a team of 6 to build PickNGo, a drive-thru click-and-collect platform that lets customers preorder and pay online, then receive purchases from mall merchants directly at their vehicles without incurring parking fees within a time-limited window.
Traditional click-and-collect can require drivers to find parking, walk to collection points, and wait in queues, adding time and parking costs. PickNGo transforms this into a drive-through handover: customers order and pay in advance, licence-plate recognition detects their arrival, and merchant staff bring their purchases directly to the vehicle while they wait. A proposed 15-minute complimentary pickup window allows customers to collect orders without incurring parking fees. Built using service-oriented architecture, PickNGo earned an 'A-' in SMU’s system design course.
StyleSnap
Collaborated in a team of 6 to produce a full-stack web application called StyleSnap - a digital closet management platform that helps users organize their wardrobe and create AI-informed outfit combinations.
StyleSnap is a wardrobe planning application designed to help users make confident outfit decisions. Many people experience uncertainty and overwhelm when choosing what to wear, especially before important events, despite having a full wardrobe. StyleSnap addresses this problem by allowing users to visualise their wardrobe, mix and match outfits digitally with or without AI, and receive trusted feedback from people who genuinely understand their personal style. This app receives an 'A' grade in a web application course at SMU.
for OAuth
Wanderlust
Built Wanderlust: a privacy-first Flutter travel companion powered by 19 modular Google ADK agents across 5 workflows to streamline itinerary research, planning, and real-time adaptation.
Travelers spend an average of 303 minutes (over five hours) consuming travel content in the 45 days before booking. Wanderlust helps to streamline this by transforming natural-language preferences into personalized, route-aware itineraries. It also provides an AI travel assistant for contextual questions, agentic discovery for applicable booking options, and AI-assisted, language-aware voice bookings - integrating Gemini Live with Twilio to place phone calls on the traveler's behalf and bridge language barriers with local staff abroad. During active trips, the app proposes adjustments when delays or route deviations disrupt the schedule. Wanderlust is privacy-first, keeping itineraries, preferences, and trip history stored locally on the user’s device.
StarNote Converter
Developed StarNote Converter - a currency exchange app with AI-based currency market analysis and prediction.
StarNote is a full-stack currency exchange web application that provides real-time exchange rates alongside AI-powered currency rate analysis and prediction. It is designed to address a key user problem: the lack of visibility into future market movements, which makes it difficult for users to plan or make informed currency exchange decisions. By combining live data with predictive insights, StarNote helps users move from reactive exchanges to more confident, forward-looking decisions.
​
I developed StarNote independently during my winter break, managing the full process from concept to implementation. The concept was grounded in the complete design thinking framework, including user research and empathy work with three defined user personas, the creation of low-fidelity and high-fidelity prototypes, and peer-conducted Nielsen heuristic evaluations to refine usability before development.
​
This app is best used using Google Chrome or Microsoft Edge.
for OAuth