🏆 Utah Tech Week AI Hackathon Winner – 1st Place out of 186 Team Members! 🏆
I’m thrilled to share that my team won 1st place at the 2025 Utah Tech Week AI Hackathon thanks to Wayne (Hanjun) Park's sacrifice and inspiration, competing against 186 teams! This marks my 7th hackathon championship, and I couldn’t be more grateful to Heavenly Father for this opportunity to make a world a better place.
Our project? TED the Bear – an AI-powered therapy robot. 🧸💡
TED the Bear could hug users, recognize speech, analyze emotions, and provide personalized counseling responses in the voices of Donald Trump, Elon Musk, or TED the Bear. Additionally, we developed a software avatar that lip-syncs in real-time to match the generated voice and language.
💻 Tech Stack & Implementation:
🔹 Frontend: React + Vite
🔹 Backend: FastAPI (CRUD functions)
🔹 Lip Sync: Gooey.AI
🔹 Face & Emotion Recognition: OpenCV + DeepFace
🔹 Speech-to-Text: Whisper AI
🔹 Database: Supabase (PostgreSQL)
🤖 Robotics Integration with AI:
🔹 Raspberry Pi 4B: Connected via Wi-Fi to Supabase for real-time data exchange
🔹 Microphone & Speaker: Captured user input and played AI-generated counseling responses
🔹 Raspberry Pi Camera: Analyzed user expressions and emotions for personalized responses
🔹 Arduino: Controlled motorized gestures with stepper motors & depth sensors (ultrasonic) to hug users within range for emotional support
Over three intense days, our team built this full-stack AI therapy robot.
⭐ Meet the Incredible Team:
👨💻 Eddy Kim – Integrated Whisper AI for speech to text, fine-tuned decibel-based input detection, helped robot internal design building, and produced and edited an amazing 2-minute demo video.
👨💻 Jonghyuk Lee – Designed the entire backend, implemented Text-to-Speech & Speech-to-Text, built facial/emotional recognition, and integrated RAG-based AI responses from OpenAI. He also set up Supabase connections and React frontend.
👨💻 Taehoon Kim – Engineered prompt optimization to determine when to use RAG vs. direct responses, solved Audio-to-Text issues, integrated AI lip sync with moviepy and wav2Lip, frontend character toggle and picture it at the center, managed the data queueing system, and integrated with each API endpoint.
👩💻 Yirang Lim – Built the infrastructure of backend system, trained AI voices to match characters like Donald Trump, Elon Musk, and TED the Bear, implemented text to speech, save those responses to DB, setup Supabase and design tables, and voice cloning using Play.ht, and built a RAG system.
🎯 My Role in the Project:
I built the full-stack development of the AI therapy robot, integrating:
✅ Raspberry Pi, Arduino, Stepper Motors, Ultrasonic Sensors
✅ Custom hugging logic & body language
✅ Real-time speech recognition & AI response system
✅ Built CRUD functions with Supabase
Our team members are actively seeking internships & full-time opportunities, and I highly recommend them to any top-tier AI or SWE teams!🚀🔥