Hello there! I’m a second-year Engineering Physics student at UBC. I build AI tools and physical prototypes.
Emergency-response scenario prototype using synthetic cellular activity to explore transport, shelter, and supply needs at Waterfront Station.
Portfolio platform for hardware engineering recruitment, with project editing, media uploads, and a searchable talent directory.
Robotic hand prototype connecting live English speech recognition to Arduino-controlled ASL hand movements.
Four-round data storytelling competition exploring survey responses through semantic maps, narrated dialogues, and conversational retrieval.
Portable wearable AI assistant (camera, mic, speaker) giving low‑latency scene descriptions & Q/A; core translated to microcontroller (C++ + Edge Impulse).
Personalized announcements + question answering by integrating Gemini with internal docs (RAG) for sales teams; selected top 3/100+ & ranked #1 in final judging by the Global Head of Partnerships with Google Canada.
Could people get home if a crossing or transit route closed? With cell-safe, our team, Five Bars 3G, explored that question around Waterfront Station using synthetic cellular activity. Rogers, UBC, and Databricks hosted the challenge.
I designed the database schema and Databricks medallion pipeline, turning 8.06 million synthetic cellular records into 437,760 minute-level activity summaries. We grouped 36 home origins into four regions and checked that regional counts reconciled to the totals.
I helped develop the team’s disaster-response scenario. We connected the activity summaries with community hubs, road closures, and transit information to explore transport, shelter, and supply needs under different conditions.
We won first place and a $3,000 team prize. The prototype made the scenario easier to inspect, while relying on explicit assumptions: synthetic session counts and home origins alone cannot establish where someone wants to go or what help they need.
Live demonstrations of JARVIS exposed microphone transcription problems, search errors, and fragile hardware mounting. Taking the wearable AI assistant to four Vancouver AI meetups gave me specific problems to work on, and people to learn from.
The first version mounted a Raspberry Pi, camera, microphone, earphones, button, and battery on a hat to describe scenes and answer spoken questions. The camera board and battery shown here are from a later, smaller prototype.
I fixed the search behavior, adapted how I used the microphone, and gathered feedback from local AI professionals. Demonstrating the assistant meant working through practical problems as well as the AI pipeline.
I worked on moving wake-word detection and speech-to-text onto a Seeed Studio XIAO ESP32S3 Sense using C++ and Edge Impulse. A $1,000 Ingenious+ grant helped fund continued development.
A few more projects and their results.
My contributionI co-founded Buildplate and worked as a full-stack developer from June through August 2026.
The resultBuilt and launched the platform, supporting 83 profiles and 66 published portfolios during my role.
My contributionI led the cross-functional team as project manager and developed the Arduino C++ codebase, hardware interface, and pulley actuation mechanism.
The resultBuilt a robotic hand prototype connecting live English speech recognition with physical ASL hand movements.
Four-round data storytelling competition exploring survey responses through semantic maps, narrated dialogues, and conversational retrieval.
The resultPlaced second overall and received the $1,000 Hall of Fame prize in the four-round competition.
My contributionI designed and built the Gemini-powered MVP through The Knowledge Society’s Google Canada challenge.
The resultGoogle Canada ranked the solution first out of more than 100 submissions.
As Community DevOps Manager and an independent contractor at the BC + AI Ecosystem from February to May 2026, I worked on Discord permissions and onboarding for a 1,000+ member community, contribution workflows, and a community project directory.
Presenting at Vancouver AI meetups gave me practice explaining technical work and listening to feedback. Earlier, I taught chess to more than 30 elementary students through the BC Junior Chess Association and maintained its website.
TKS, 2023–2025 → Langara Engineering Transfer, 2025–2026 → UBC Engineering Physics, 2026–2030 (expected).
I’m interested in taking this mix of software and physical engineering further into robotics and quantum computing.
Spelling Garden began with my younger brother’s spelling bee. Other small builds explore job listings and Vancouver transit.
Spelling Garden started as a way to help my younger brother prepare for his spelling bee. It pairs spoken words with spaced repetition for daily practice.
Vancouver Transit Tracker turns transit schedules into an animated map of SkyTrain, buses, and SeaBus. Positions are schedule estimates, rather than live vehicle measurements.
Scoped brings UBC SCOPE listings into a searchable, filterable workspace with saved positions and a globe. Semantic search runs locally in the browser.
Hackathons give me a reason to cross disciplines, build with friends, and test an idea under a tight deadline. A few experiments beyond the selected work above:
Deaf Leppard · Hack the North, 2026. Our team combined speaker-attributed captions in AR glasses with directional haptic alerts, working across computer vision, embedded software, and circuitry.
Genimo · nwHacks, 2025. Inspired by 3Blue1Brown, we connected Gemini to Manim to turn math and physics questions into educational animations. I worked on prompts and the generation pipeline.
Pet Maker · StormHacks, 2026. Eric Chen and I turned pet photos into rigged 3D companions that respond to hand tracking, petting, and fetch. Making gestures and contact feel consistent was the interesting challenge.
If you’re working on AI, software, or hardware, or have a question about something here, I’d like to hear from you.
Hello there! I’m a second-year Engineering Physics student at UBC. I build AI tools and physical prototypes.
Engineering
