At BagelHacks II, we built and pitched Ledger AI, an AI-powered expense intelligence platform for SMBs. It replaces slow, manual finance workflows with a conversational interface powered by NLP and LLMs, allowing users to query data in plain English. Behind the scenes, real-time anomaly detection identifies fraud patterns and enforces policy compliance using contextual reasoning. Built on scalable microservices with API integrations and live data pipelines, the platform also delivers explainable AI recommendations and automated, CFO-ready reports—reducing manual work, accelerating approvals, and driving up to 41x ROI.
During this challenge, our team developed an AI agent designed to help companies identify in less than 30 seconds the optimal building location based on their clients needs. By leveraging data and intelligent analysis, our solution evaluates key factors such as parking availability, foot traffic, street context, and accessibility to support smarter, data-driven decisions. Grateful to EY Montréal and Databricks for organizing such an inspiring event, and to everyone who contributed to this experience.
Built during a 48-hour game jam by a team of two, Jiejun and me, this project challenged us to create and launch a fully playable game from scratch in a Shrek-inspired universe. Our game, Shrek: 7 Layers Onion, went on to win Best Themed Game, reflecting both creative execution and strong technical problem-solving under pressure. Throughout the jam, we designed different scenarios and gameplay moments that matched the theme while keeping the experience fun and coherent. No AI was used in the creation process. Link: https://www.roblox.com/games/119841885626750/Shrek-7-Layers-Onion