Build for Odisha

Teams of IIT students, alumni and young founders take on real problems set by government and industry, with mentors along the way and a jury at the summit.

Five Steps to the Final Demo

  1. Sign In

    Open your summit network account: from your ticket link, or with your email.

  2. Form a Team

    Create a team and share its join code, or join a team with a code from its lead.

  3. Pick a Problem

    Choose one of the problem statements below. You can change it until the final submission.

  4. Submit at Each Milestone

    Hand in a link (repository, deck or video) by each date. Mentors comment on your work.

  5. Demo to the Jury

    The jury scores the final submissions and the best teams present at the summit.

7 Problem Statements to Solve

  • GovTech · Proposed: a state procurement department (to be confirmed)

    Read a tender and check every bid against it

    Evaluation committees read hundreds of pages of tenders and bids by hand. Pull the eligibility conditions out of a tender, then check each bidder’s documents against them, and show the page every answer came from.

    More Detail

    Tenders and bids arrive as PDFs and scanned pages, some in Odia. A strong entry handles scans, flags what it is unsure of instead of guessing, and leaves the decision to the committee. Practise on published tenders from the state e-procurement portal; never use real bidder documents.

    Data and Starter Material ↗

  • Disaster Resilience · Proposed: the state disaster management authority (to be confirmed)

    Get cyclone warnings to every household, in Odia, in time

    Odisha’s cyclone shelters have saved thousands of lives, but the last mile still depends on volunteers going door to door. Turn official bulletins into plain Odia messages for each village and track which families have reached a shelter.

    More Detail

    Assume power and mobile data fail first. Think about basic phones, shared phones, people who cannot read, and volunteers working offline who sync later. Use public weather bulletins and published shelter locations to build and test.

    Data and Starter Material ↗

  • Mining & Metals · Proposed: A Mining Partner (to be confirmed)

    Spot unsafe acts in open-cast mines from camera feeds

    Haul roads and benches are watched by cameras, but nobody can watch them all. Flag people too close to moving machines, missing helmets and vehicles in no-go zones.

    More Detail

    Registered teams get a set of anonymised clips. A good entry runs on a modest laptop or an edge box at the pit, explains each alert, and keeps false alarms low enough that supervisors trust it.

  • Energy · Proposed: a power distribution company (to be confirmed)

    Forecast rooftop solar on each feeder, a day ahead

    Rooftop solar changes how much power each feeder needs. Forecast the next day hour by hour so the grid can plan purchases and avoid overloads.

    More Detail

    Anonymised feeder data for a year is shared with registered teams; public satellite irradiance data helps you start before then. Open models and honest error measures score well.

    Data and Starter Material ↗

  • Deep Tech · Proposed: A Semiconductor Partner (to be confirmed)

    Find the pattern behind defects on a silicon wafer

    Odisha is building its first chip plants. Wafer test maps show where dies fail; the pattern (a ring, a scratch, a corner) points to the faulty step. Classify the patterns and flag new ones nobody has labelled.

    More Detail

    Start with the public WM-811K wafer map set. Most maps have no defect pattern and some patterns are rare, so measure how you do on the rare ones, not just overall accuracy. Explain why each map was classified the way it was.

    Data and Starter Material ↗

  • Health · Proposed: the state health mission (to be confirmed)

    Spot High-Risk Pregnancies Early in Remote Blocks

    Health workers in hilly and tribal blocks see each mother only a few times. Give them a simple tool that uses what they already record to flag who needs a doctor’s visit, and when.

    More Detail

    It must work offline on a basic Android phone, in Odia and simple pictures, and fit into a visit without adding paperwork. Build on public statistics from the National Family Health Survey and on synthetic records; no real patient data will be shared during the hackathon.

  • Smart Cities · Proposed: the district administration, Puri (to be confirmed)

    Warn before crowds get dangerous on Puri’s Grand Road

    During the Rath Yatra, over a million people pack a few kilometres of road. Estimate crowd density from camera views and predict, a few minutes ahead, where it will cross a safe limit.

    More Detail

    Counts must not identify anyone: no faces, no tracking of individuals. Show controllers a map they can read at a glance and suggest which barricade or lane to open. Public crowd-counting research sets are a good start until anonymised footage is shared.

Key Dates

  1. Registration Closes

    Create your team (2 to 5 people), share the join code with your members and pick a problem statement. Students, alumni and young founders from any IIT can take part.

  2. Idea Round

    A two-page note or a deck of up to six slides, shared as a link: the problem as you understand it, your approach, the data you will use and how you will know it works.

  3. Mentors Assigned

    Every team with an idea note gets a mentor from industry or academia. Use the notes on your team page to ask questions.

  4. Working Prototype

    Link to your public repository and a 2-minute screen recording on YouTube or Vimeo (unlisted is fine). Mentors comment on it within two days.

  5. Final Submission

    Final repository, a demo video of up to 3 minutes and a short note on what works, what does not and what it would take to pilot. The jury scores this one; the top six teams are told on 8 January.

  6. Live Finale at the Summit

    The top six teams demo in the Main Hall: 5 minutes each, then 3 minutes of questions. Winners are announced at the valedictory.

Judging: How the Jury Scores

  • Impact30%

    How much it would help the people who set the problem, and how many people, if it were used tomorrow.

  • Working Solution30%

    It runs on real or realistic data, the demo shows it, and the results are measured honestly.

  • Ready to Pilot20%

    Cost, data needs, Odia and offline use, privacy and the people who would run it have been thought through.

  • Technical Depth10%

    Sound engineering and a fresh approach, not a thin wrapper around something that already exists.

  • Presentation10%

    A clear demo and straight answers to the jury’s questions.

Each juror scores every final submission from 0 to 10 on each point. The weighted scores are averaged across jurors.