Neurotech Internals

Overview

Running a student neurotech club meant a lot of spreadsheets: who is a member, who showed up, who owes dues. Internals replaces them with one Svelte app on AWS Serverless.

Attendance is the part worth describing. An admin creates an event with a name, time and a latitude and longitude. Members upload a photo of themselves at the event, and the app reads the EXIF capture location out of the image to confirm they were actually there. The extraction happens client-side and the photo goes directly to S3 from the browser, so the API never touches image bytes — traffic spikes hard during an event and barely exists otherwise, and this keeps the spike off the functions entirely. A Lambda then works through an SQS queue comparing photo location to event location, currently against a 1 km threshold.

Anything ambiguous lands in a review queue instead of being guessed at. Admins page through the flagged entries with presigned S3 URLs and approve or reject each one, which matters because the honest failure mode of EXIF-based attendance is a stripped or missing location tag, not a forged one.

The stack is deliberately cheap: Svelte and Hono for a small bundle, DynamoDB because the free tier covers a club's entire dataset, S3 for photos. Dues and merch tracking are the next thing to automate.