Spinsight AI

A full-stack solution for storing, categorizing, and analyzing pole dancing media using AWS Bedrock AI for pose recognition and technique analysis.

Comment

Inspiration

The inspiration for Spinsight AI came from my love for pole dance. As a pole dancer, I know how important it is to receive constructive feedback, track progress, and celebrate every milestone.

What it does

Spinsight AI is a web application that allows pole dancers to upload, organize, and analyze their training photos. Using AI, the app provides detailed feedback on form, technique, and safety for each uploaded image. Dancers can track their progress, revisit their best moments, and receive personalized suggestions for improvement.

How we built it

  • Frontend: Built with Angular for a modern, responsive user experience.
  • Backend: Utilizes AWS Lambda, DynamoDB, and S3 for secure, scalable media storage and retrieval.
  • AI Integration: Leverages Amazon Bedrock’s Claude Vision to analyze images and generate feedback.
  • Infrastructure as Code: Managed with Terraform for reproducible, cloud-native deployments.
  • OpenAPI: Used for clear API contracts and auto-generated frontend models.

Challenges we ran into

  • AI Integration: Integrating Bedrock, since it was the first time using this service. Also choosing the right model was a challenge, because it is a field where I am not experienced at.
  • Data Serialization: Handling the conversion between Python floats and DynamoDB Decimals, was a subtle but critical challenge.
  • User Experience: Designing a UI took several iterations and lots of user feedback.

Accomplishments that we're proud of

  • To have integrated Bedrock for the first time in an application.
  • A beautiful, intuitive gallery for organizing and revisiting progress.
  • Robust backend that scales with user needs and keeps data safe.
  • Automated model generation and infrastructure, making the project easy to maintain and extend.

What we learned

  • The importance of clear data contracts (OpenAPI) for frontend-backend communication.
  • That even small technical details (like number serialization) can have a big impact on reliability.
  • Integration of lambda with Bedrock.
  • Develop the application in several iterations improving backend and frontend at the same time.

What's next for Spinsight AI

  • Video Analysis: Expanding AI feedback to cover videos, not just images.
  • Mobile Experience: Create a progressive web application
  • Security: Since I developed the application in for the hackathon, I will need to secure it with login, Cognito and other security configurations.

Built With

Share this project:

Updates