Spatial Concept Lab

Inspiration

Studying abstract concepts you do not understand on your own is hard. Text, videos, and diagrams can help, but they often still feel separate from the learner's thinking process.

Spatial Concept Lab was inspired by the idea that abstract concepts become easier to understand when people can visualize them, interact with them, and leave shared notes for the next learner. Instead of treating a lesson as a flat page, the project treats the concept as a shared room.

What It Does

Spatial Concept Lab is a 3D social learning prototype. A learner enters a private demo, builds through an abstract concept, chooses a display name, and leaves a named note directly on a 3D object.

The note becomes a persistent shared marker. It includes the user's display name, marker text, room ID, concept ID, timestamp, and 3D coordinates. The same saved marker can be loaded from another device or browser session after refresh.

How We Built It

The frontend is built with Vite, React, TypeScript, and Three.js, then deployed on Vercel.

The backend uses Vercel Functions as the server-side boundary. The browser calls /api/markers, and the function validates marker data before writing to AWS.

For persistence, the app uses Amazon DynamoDB. The deployed Vercel Function accesses AWS through Vercel OIDC and an AWS IAM role, so AWS credentials are not exposed in the browser.

Challenges We Ran Into

The biggest challenge was deciding what should be persisted. A 3D social learning app could become expensive quickly if every camera movement, pointer event, or avatar action were synchronized in realtime.

The prototype solves this by starting with async multiplayer. It saves durable learning artifacts: named notes, 3D positions, concept IDs, room IDs, and timestamps. Manual refresh is an intentional cost-control tradeoff.

Another challenge was runtime authentication between Vercel and AWS. The browser cannot safely hold AWS credentials, so the project uses a Vercel Function and AWS IAM role access through Vercel OIDC.

There was also a schedule challenge around v0. The supplier credit deadline closed on June 26, 2026, and I could not finish the submission workflow inside v0. I confirmed that the submission can still use a Vercel-deployed frontend as long as the required AWS database usage is shown.

Accomplishments That We're Proud Of

We built and deployed a working cloud-backed prototype in the final hackathon window.

The app has a live Vercel frontend, a private demo access gate, a 3D learning interaction, named spatial markers, a Vercel Function API boundary, and DynamoDB persistence.

The strongest proof is that markers are not only local browser state. They are saved through the production API and can load back in a fresh session.

What We Learned

We learned that "multiplayer" does not have to start with expensive realtime infrastructure. For a learning product, async shared artifacts can prove the social value first.

We also learned the difference between developer access and runtime access. Developer CLI access is useful for setup and verification, but the deployed app needs its own secure path. Vercel OIDC plus AWS IAM made it possible for the function to use temporary AWS access without browser-side secrets.

What's Next For Spatial Concept Lab

The next step is improving the actual visualization process for abstract concepts. Right now the prototype proves the backend loop: visual lesson, named marker, serverless validation, and DynamoDB persistence.

Future work would make the concept-building experience richer, add more learning templates, improve marker placement controls, add optional low-frequency polling, and create a guided judge/demo mode inside the app.

Built With

• amazon
• dynamodb
• for
• functions
• iam
• javascript
• node.js
• oidc
• react
• sdk
• three.js
• typescript
• v0
• vite