BarClip
Training videos are full of dead time from setting up and stopping the phone, and after a workout, manually trimming each one is the last thing you feel like doing. BarClip automatically detects and trims the lift from any training video, and can batch-process the full session into one edited clip ready for social media.
- Reduces time for batch video processing from 1.5 hours to 15 minutes
- Trained a custom YOLOv8 computer vision model to detect weightlifting plates and built an algorithm to translate detections into motion tracking
- Developed a cloud-based MVP with a React front end and a containerized Azure Function (Docker, Azure Container Apps) for video processing, then pivoted to a local-processing architecture to reduce cloud costs and improve performance.
- Set up a CI/CD pipeline with GitHub Actions to build and deploy the .NET MAUI iOS app to TestFlight, enabling on-device testing without Mac hardware
- Repurposed the original web app's API, SQL Server database (EF Core), and Entra External ID authentication into the mobile app's backend once local processing eliminated the need for cloud video handling
C#
.NET
MAUI
Computer Vision
Azure
React
SelfTalk
Professional startup work on a React Native mobile app with a .NET backend and Azure cloud architecture for personal self-reflection and mindfulness practices.
- Architected an audio processing microservice using Azure Functions, Event Grid, and Blob Storage, cutting perceived API latency from 4.7 seconds to 100ms
- Refactored the Playlist API from full-object to partial updates, reducing payload size by 80% for common operations
- Designed and implemented Azure B2C authentication across frontend and backend to protect user-uploaded audio content
- Integrated third-party voice isolation to improve audio processing stability and reduce retries during testing
- Built cross-platform mobile app with React Native, Expo, and TypeScript
- Collaborated with a senior engineer via GitHub branching, PRs, and code reviews
React Native
C#
ASP.NET Core
Azure