For a small college hackathon, plan three to five judges per project, then add two or three standby judges in case someone drops out. The total panel size depends on your number of teams, the minutes you can give each team, and how many judges must see each project. A small event may need only a handful of people. MLH's worked example needs 18 judges for 175 projects, so a large event can need many more.
Why the sources give different numbers
The guides we found do not agree on one headline figure. They also measure different things. ScoreJudge says five to ten judges is typical for student and corporate hackathons, and that large public events sometimes use fifteen or more [3]. Another guide says three to five judges works best for a panel [5]. MLH gives no fixed headcount. It gives a formula that scales with the number of projects [1].
The difference comes mostly from project count and available time. A panel that works for a small field may be too small for a larger one in the same window. We found no source written specifically for India. Treat these figures as general hackathon practice, not rules for your campus.
The formula behind the numbers
MLH's judging-plan guide works backwards from four inputs: the number of submitted projects, the judges assigned to each project, the minutes each project takes, and the total judging time available [1]. Its recommended inputs are three judges per project and four minutes per project. Those four minutes break down into two for the demo, one for questions and scoring, and one for judges to move between rooms [1].
In MLH's worked example, 175 submitted projects and two hours of judging need 18 judges [1]. You can run the same logic with your own numbers:
| Input | What it means | Why it matters |
|---|---|---|
| Submitted projects | Teams that will be judged | Drives the total workload |
| Judges per project | MLH recommends 3 | Each project needs several independent views |
| Minutes per project | MLH uses 4 (2 demo, 1 questions and scoring, 1 travel) | Sets how many projects one judge can see |
| Total judging time | The window you can actually use | Caps how many judges you need to finish |
The formula does not give the number of judges directly. It gives the total judge-slots you need. You then divide those slots among people who are actually available.
Judges per project and the total panel
The total panel and the judges per project are different numbers. A panel of eight judges does not mean each team sees all eight. MLH says each project should be seen by at least three different judges. Judge groups should be staggered so the same teams are not always judged together [1].
An AI-tools vendor guide puts the practical minimum at three judges per project, which reduces the effect of one person's bias. It recommends five when the event covers both technical and business criteria. It also warns that more than five judges per project usually slows things down unless scoring is split by category [4]. Each extra judge adds a seat to fill, a briefing to give and a score to reconcile. Past the point the arithmetic needs, those costs grow faster than the benefit.
Devpost notes that one judging panel is the most common setup. When an event uses several category panels, each panel should have at least three diverse members [6]. For a student hackathon with one set of criteria, one panel is usually simpler to run.
Rounds, finalists and category panels
MLH says judging usually takes two to three hours, and many events run a single round. Some events pick finalists for the closing ceremony, which adds a second, smaller stage [1].
Two examples show how the structure can change. The 2025 Harvard rare disease hackathon used preliminary rooms with two judges each, then a final round with a full panel of 12 judges [7]. That split suits events where the first round must cover many teams quickly and the final round needs more weight. A small club hackathon may not need it.
If your event has separate tracks, such as a technical prize and a social-impact prize, consider one panel per track rather than one large panel. Devpost's minimum of three members applies to each panel [6].
Standby judges and recruiting
MLH advises recruiting two or three extra judges as a standby list. A single drop-out can disrupt the whole plan, and the standbys protect the schedule [2]. It also suggests starting recruitment in the last two months before the event [2].
For a college event, ask every judge to confirm in writing. Name your standbys before the day. Check that they can reach the venue within the judging window. Our guide on how to make a budget for a college fest shows how to list every expense head before you spend. That helps you cost judging alongside the rest of the event.
A worked example
Suppose a college hackathon has 60 teams. Using MLH's three-judges-per-project rule, that is 180 judge-evaluations. At four minutes each, the total is 720 judge-minutes. Spread across a two-hour judging window, which is 120 minutes, you need six judges working at the same time. Add two standbys and you have eight people on the list.
The numbers are hypothetical, but the method holds. A different team count or window changes the answer directly, so rerun the arithmetic with your own figures before you recruit. For a sense of scale, a judging-software vendor's illustrative expo used about 60 projects, eight judges plus two standbys, three judges per project and a 90-minute window [8]. The vendor presents it as an example, not a standard. Use it only as a sanity check.
Our view
Our view: for a student hackathon, start from the three-judges-per-project rule. Run the MLH arithmetic with your real team count and window before you recruit anyone. Put your effort into the rubric and the judge briefing, since the rubric is what keeps scores comparable across judges. Extra judges beyond what the arithmetic needs mostly create scheduling problems.
Limitation
The guidance here is general, not India-specific. The sources are guides and vendor pages, and we found none written for Indian college events. The MLH formula assumes four minutes per project including travel. That may be too short if your rooms are spread across a large campus. Unifesto's Sprint product supports judge panels and weighted rounds, but we have no data on how many judges Indian college hackathons actually use. This post cannot tell you what is normal at your college.
FAQ
Is three judges per project enough?
For most small events, yes. MLH says each project should be seen by at least three different judges, and several guides give three as the practical minimum [1][4].
Should every judge see every team?
No. Stagger the judge groups so the same teams are not always judged together. Each project only needs about three judges [1].
How many standby judges should I recruit?
MLH advises two or three extra judges as a standby list, because any drop-out can disrupt the whole plan [2].
Can I use one panel for a whole hackathon?
Yes. Devpost says one judging panel is the most common setup [6]. Use separate category panels only when you have distinct tracks, and give each at least three members.
If your team is still planning the hackathon itself, our guide on how to organise a college hackathon covers format, rooms and rubric. For the people running check-in and help desks during judging, how many volunteers a college fest needs gives rough staffing ratios. If you run a Sprint hackathon on Unifesto, Sprint supports judge panels, weighted rounds and a live leaderboard.
Sources
- Judging Plan: https://guide.mlh.com/general-information/judging-and-submissions/judging-plan
- Judges Communication and Recruiting: https://guide.mlh.com/general-information/judging-and-submissions/judges-communication-and-recruiting
- Judging Software for Hackathons: https://scorejudge.com/judging-software-for-hackathons/
- How to judge hackathon: The complete guide: https://ai-beavers.com/blog/how-to-judge-hackathon-complete-guide
- How to Judge a Competition Fairly: Best Practices for Event Organizers: https://scorejudge.com/blog/posts/how-to-judge-a-competition-fairly/
- Judging & public voting - Devpost.com Help Center: https://help.devpost.com/article/64-judging-public-voting
- Judges Guide (View), HackRare: A Rare Disease Hackathon: https://www.hackrare.org/judges-guide-view
- Free Hackathon Judging Scoresheet Tool You Can Build: https://rationalgo.ai/resources/app-builder/community-tools/hackathon-judging-scoresheet-tool