The hardest part of a first hackathon is not building. It is spending five hours arguing about what to build and then picking something you cannot possibly finish. Here are forty starting points, sorted by how realistic each one is in a weekend.
Read this before you pick
Three rules that matter more than the list itself:
- Pick something you personally need. You already know what it should do and you can tell instantly whether it works. This one heuristic beats every brainstorming technique.
- Then make it smaller. Whatever scope feels right, halve it. You can always add more at hour 30; you cannot recover a project that was too big at hour 30.
- Originality of approach beats originality of topic. A well-executed attendance tracker scores better than a vague, half-built AI platform. Judges have seen the ambitious version fail hundreds of times.
Genuinely beginner (finishable in a weekend, first hackathon)
These need one screen or one core loop. Judges score them well when they are polished, because a small thing that works demonstrates more than a big thing that does not.
- Attendance tracker that calculates how many classes you can still skip
- Timetable app that tells you only what is next, and nothing else
- Lost and found board for your campus with photo upload
- Mess menu app with a one-tap rating per meal
- Hostel room-cleaning or chore rota with reminders
- Split-the-bill calculator built for hostel groups, not restaurants
- Assignment deadline tracker that colour-codes by panic level
- A "what should I eat" randomiser for the canteen
- Simple habit tracker with streaks, no account required
- Textbook and notes swap board for juniors and seniors
One step up (you will need to learn one new thing)
These involve an API, a database or a small amount of data processing. Very achievable inside one round with mentor help, and they show more range.
- Turn lecture slides into practice questions automatically
- Past-papers library with proper search across subjects
- Carpool matcher for students on the same daily route
- Anonymous peer-support board with basic moderation
- Campus event calendar aggregated from club announcements
- Study-group finder that matches by subject and free hours
- Medicine or hydration reminder that works offline
- Bus and local transport timing board for your specific city
- Resale marketplace for cycles, calculators and lab coats
- A tool that flags accessibility problems on your college website
Ambitious (only with someone experienced on the team)
Worth attempting if your team already has someone who has built something before. Attempt one of these solo as a first-timer and you will most likely not submit.
- Summarise a long PDF into a revision sheet with page citations
- Voice-to-text notes that handle Indian-accented English properly
- Civic issue reporter that maps potholes and waterlogging from photos
- Regional-language translator for government and college forms
- Waste segregation guide that identifies items from a photo
- Marks and CGPA predictor from your own historical data
- Plagiarism-style similarity checker for student assignments
- Real-time collaborative whiteboard for remote study groups
- Local-language voice interface for a college helpdesk
- Crowd-sourced review platform for college labs and electives
Non-obvious categories judges remember
Everyone builds an app. These stand out simply because almost nobody in a first-year cohort attempts them.
- A browser game with one genuinely unusual mechanic
- A developer tool that removes one annoying step from your own workflow
- A data visualisation that reveals something surprising in public data
- A browser extension that fixes something specific and irritating
- A physical-world simulation — traffic, queues, crowd flow
- An accessibility tool for a specific disability, built with real input
- A tiny hardware simulation with a web front end
- A bot that automates something your college society does by hand
- A teaching tool that explains one hard concept interactively
- An archive or documentation project for something local and undocumented
How judges actually score these
Understanding this changes which idea you pick.
- Innovation (~30%) — not "is this topic new" but "did you approach it in a way we have not seen twenty times today". A common topic with one genuinely clever decision scores well.
- Technical execution (~30%) — does it run, is the code followable. Note that this is judged relative to your track, so a first-year is not measured against a final-year build.
- Impact (~20%) — can you name a specific person who would use this and say why. "Students" is not a person. "A second-year in a hostel who keeps missing the laundry pickup" is.
- Presentation (~20%) — your demo video and README. This is the easiest 20% on the board and the one first-timers most often leave on the table.
Ideas that consistently score badly
Not because they are bad ideas, but because they are hard to demonstrate well in three minutes:
- A thin wrapper around a chat API. Judges see dozens. If your project is a text box that forwards to a model, the model did the work, not you.
- A social network for X. Impossible to demo meaningfully with no users.
- A dashboard with no real data behind it. Placeholder numbers are instantly visible and read as unfinished.
- Anything medical that claims to diagnose. Marked down heavily on responsibility grounds at most events, ours included.
- A clone of an existing product with nothing added. Fine as practice, weak as a submission.
Still cannot decide? Use this
Set a 30-minute timer at kickoff. Each team member writes down three things that annoyed them in the last month — anything, not just technical. Put all of them on one list, cross out anything that needs data you do not have or hardware you cannot get, and pick the one that the most people in the team recognise. Start building at minute 31.
A decent idea started at hour 1 beats a great idea started at hour 6. This is consistently true and consistently ignored.