Poster Judging Platform

My sister hosts conferences pretty often, however, there is one problem: the judging process. Currently, she hands out sheets of paper to the judges for them to fill out and return. This gets tedious really fast, sometimes there are judges that add up their totals wrong, sometimes they don't return their sheet (and she has to chase them down). All in all, it's a mess.

A few days ago, she asked me if I could build her a platform to make the judging process easier, and I couldn't say no to the opportunity of learning something new.

What it does

This platform allows judges to log in (via an access code the organizer generates and hands to them on a printed card, no one wants to make a pesky account for a single event). After logging in, judges can submit their scores for each poster they were assigned (the organizer can assign posters), they submit it and ta-dah! The posters are all rated, and the organizer isn't chasing down judges for their scores, correcting their math or doing a mad minute to tally up the results.

How it works

This is a Next.js 16 app that replaces paper judging sheets with a web platform. It has a few significant parts to it.

Main Pages

There are two main pages in this app, the organizer page (only organizers can see this page, and only for their own events), and the judge page. Judges receive a unique access code from the organizer, which they can use to log in and start judging presentations/posters. Each judge has specific posters assigned to them, and once their scores are submitted, the organizer can view the results and see which posters received which scores from who.

Under the hood

Under the hood, this app uses Next.js and the backend is powered by a Postgres database (hosted on Supabase). I've written a breakdown on some of the components of the app and how they work:

Logging in: Each judge gets an 8-character code like K7M2-Q9XR. The characters come from an alphabet with all the lookalikes removed (no 0/O, no 1/I/L), so nobody squinting at a card in a loud hall types the wrong thing (and then starts getting mad at the organizer). The database never stores the code itself, only a hash of it, so the organizer sees each code exactly once, when they're generated. When a judge types their code in, it gets hashed, finds the matching judge, and gives their phone a signed cookie that lasts the event day. To stop people from guessing codes, a code gets locked out after 10 wrong tries in 15 minutes. That limit is tied to the code. Organizers log in separately, with a normal email and password through Supabase Auth, and for them, they'll get an account that they can use.

Scoring: The organizer sets up the rubric, and each criterion has its own maximum and its own weight. To get a total, each score becomes a fraction of its maximum, gets multiplied by its weight, and the whole thing is averaged into a percentage. That means a 1–5 criterion and a 1–10 criterion can live on the same sheet. For example, Clarity 4/5 (weight 1) and Research 7/10 (weight 2) works out to (0.8 × 1 + 0.7 × 2) / 3 ≈ 73%. This is set to change depending on user feedback and organizer needs.

Assigning posters: Currently, posters are assigned automatically to judges, however, this is a work in progress. Specifically, I'm looking to add a feature where organizers can tag posters and assign them to judges based on these tags. This is important so that judges can be assigned to posters that are relevant to their area of expertise.

Accounting for bad wifi: Oftentimes bad wifi can be a problem in large venues (not everyone has LTE/5G), so the scoring form assumes the connection will drop. Every tap saves the sheet on the phone itself, so nothing gets lost in case the phone dies or you exit out of the site. If a submit fails, it goes into an outbox and the judge just moves on to the next poster. The app retries every 20 seconds and as soon as the phone is back online. Sending the same sheet twice is harmless, because the database keeps exactly one sheet per judge per poster and a resend just overwrites it instead of counting it twice.

Lockdown defense: The browser never talks to the database directly. Every table in Supabase is locked so the public key can't read or write anything, and all reads and writes go through the server, which checks who you are first. Events belong to organizations, so two people running two different conferences on the same deployment can't see each other's posters, judges or results.

Next Steps

Since the project is still a work in progress, there are a few tweaks I plan on making after going back and forth with my sister! More updates here soon! Expect more here before January 2027.