PROBLEM STATEMENT BANK

Internal Hackathon
2026 Edition

26 real-world challenges across 10 domains. One weekend to build something that matters.

CLIMATE & CITIES SPACE & DEEP TECH GOVTECH FINTECH & SOCIAL DOCUMENT AI
IMPACT

LEGALTECH EDTECH PUBLIC SAFETY GST & MSME TECH PRIVACY & AI
GOVERNANCE

Prepared for participating teams | Confidential — internal circulation only

WHAT'S INSIDE

Contents
01 Urban Heat Mitigation via AI/ML.......................................................................................... CLIMATE & CITIES

02 Forecasting Energetic Particle Radiation for ISRO Satellites........................................SPACE & DEEP TECH

03 AI-Powered Governance for Citizen Service Delivery......................................................................GOVTECH

04 Instant Climate-Resilience Insurance for Gig Workers.......................................FINTECH & SOCIAL IMPACT

05 Intelligent Document Digitization & Classification...................................................................DOCUMENT AI

06 The Record That Everyone Holds and Nobody Trusts..................................................................LEGALTECH

07 The Tutorial That Taught Nothing.......................................................................................................... EDTECH
08 The Gap Between the Train and the Cab................................................................................. PUBLIC SAFETY

09 The CA in Your Pocket That Does Not Exist........................................................................GST & MSME TECH

10 The Right to Be Forgotten That Goes Nowhere.................................................PRIVACY & AI GOVERNANCE

11 Agri Smart — Smart Digital Farming Platform....................................................................CLIMATE & CITIES

12 Crowdsourced Accessibility Mapping Platform..................................................................CLIMATE & CITIES

13 Ward-Level Urban Risk Mapper............................................................................................ CLIMATE & CITIES

14 Smart Predict — AI-Based Prediction System.................................................................SPACE & DEEP TECH

15 Medicine Shortage & Price Monitoring System................................................................................GOVTECH

16 AYUSH Startup Registration Platform................................................................................................. GOVTECH
17 Citizen Grievance Triage Engine.......................................................................................................... GOVTECH
18 Real-Time Sign Language Translator.....................................................................FINTECH & SOCIAL IMPACT

19 Verified Local-Connect Platform for Tourism......................................................FINTECH & SOCIAL IMPACT

20 Digital Marketplace for Ethical Waste Aggregation............................................FINTECH & SOCIAL IMPACT

21 Bank-Loan Due-Diligence Assistant.......................................................................FINTECH & SOCIAL IMPACT

22 Case Document Consistency Checker............................................................................................. LEGALTECH

23 Parental Screen-Time & Digital Wellbeing Dashboard.......................................................................EDTECH
24 Student Learning-Gap Resolver.............................................................................................................. EDTECH
25 Multi-Source Incident Verification System...............................................................................PUBLIC SAFETY

26 Privacy-Safe AI Audit Layer for Public Sector.....................................................PRIVACY & AI GOVERNANCE

OR
Open Problem Statement (Bring your own Problem Statement)
 CLIMATE & CITIES

01 Optimizing Urban Heat Mitigation & Cooling

Strategies via AI/ML

Develop a geospatial AI/ML system, backed by physics-informed decision-making, to identify urban heat stress
hotspots, quantify the key drivers of urban heating, and generate optimized, scenario-based cooling interventions.

Objective
 Identify hotspots — generate heat stress maps using satellite and meteorological data.
 Analyze drivers — quantify the influence of land use/land cover, urban morphology, vegetation, and
atmospheric conditions.
 Model heat dynamics — establish LST relationships using physics-informed ML.
 Optimize cooling scenarios — simulate interventions such as greening, cool roofs, albedo changes, and water
bodies.

Key Datasets
 Remote sensing: Landsat 8 & ECOSTRESS LST, Sentinel-2/Landsat LULC
 Meteorological: ERA5 & CPCB atmospheric variables
 Urban form: OpenStreetMap, GHSL, UT-GLOBUS (optional)
 Optional modeling tools: SOLWEIG, InVEST

Expected Outcomes
 Heat stress maps identifying urban hotspots
 Quantitative assessment of key heat drivers
 Validated AI/ML model of urban heat dynamics
 Optimal intervention strategy — type, spatial placement, and estimated °C reduction
 SPACE & DEEP TECH

02 Forecasting Energetic Particle Radiation for ISRO's

Geostationary Satellites

Develop and demonstrate an algorithm to predict energetic electron fluxes at geostationary orbit — with at least 30–
45 minutes of lead time, plus 6-hour and 12-hour ahead forecasts.

Objective
 Read, process, and visualize electron flux and solar wind data archived in CDF format
 Identify a suitable AI/ML algorithm for time-series forecasting
 Develop, fine-tune, and optimize the algorithm across training, validation, and testing
 Demonstrate and visualize forecast outputs and their accuracy

Dataset Required
 GOES series >2 MeV electron fluxes (CDF), 11 years
 Wind spacecraft solar wind parameters — speed, IMF, density — 11 years
 ISRO GRASP/GSAT payload electron fluxes for 1–2 years, for validation at Indian longitude

Evaluation Parameters
 Grasp of solar wind and radiation belt fundamentals
 Correctness of data reading and visualization
 Quality of algorithm identification and optimization
 Accuracy of predicted fluxes at geostationary orbit
 GOVTECH

03 AI-Powered Governance: Transforming Citizen

Service Delivery

Build a secure, AI-driven governance platform that turns raw government data into predictive, actionable intelligence
— resolving citizen service needs before they escalate.

The Challenge
The Government of Maharashtra holds vast multi-sectoral datasets across health, infrastructure, and public safety —
but they remain siloed, reactive, and underused for proactive decisions, causing delayed services and unpredictable
resource allocation.

Solution Capabilities
 Predictive AI models — forecast service demand, resource bottlenecks, and public health/infrastructure risks.
 Dynamic prioritization engine — auto-route and triage urgent requests using predictive scores and live
feedback.
 Privacy & security framework — compliance-by-design across all state and national data governance standards.
 Actionable dashboards — clear, visual recommendations for department heads.
 Governance-as-a-Service — transparent citizen oversight that builds measurable public trust.
 FINTECH & SOCIAL IMPACT

04 Instant Climate-Resilience Insurance for Gig

Workers

India's 15M+ gig and freelance workers have zero financial safety net. When floods, cyclones, heatwaves, or air-
quality emergencies strike, they lose income instantly — with no employer benefit, paid leave, or insurance to fall
back on.

Why Traditional Insurance Fails

 Slow, paperwork-heavy claims workers can't afford to wait for
 Coverage that's unaffordable or simply doesn't exist for daily-wage workers
 No real-time link between a disaster event and compensation
 No path to build financial credibility or access future credit

The Challenge
How might we build a system that automatically detects when a gig worker's livelihood is disrupted, and delivers a
fair, verified payout within minutes — no paperwork, no claims agents, no delays?

What a Solution Should Address

 Real-time, hyper-local (city/pin-code) disruption detection — weather, AQI, traffic, platform outages
 Automated, zero-paperwork verification of impact
 Instant, low-cost micro-insurance payouts suited to daily-wage income
 Fraud prevention, so payouts reach only genuinely affected workers
 A scalable, API-first architecture that plugs into any gig platform
 DOCUMENT AI

05 Intelligent Document Digitization & Classification

Government and institutional archives hold decades of paper records — application forms, ledgers, certificates,
correspondence — that are slow to search, easy to lose, and costly to maintain. Develop an AI system that converts
these physical records into structured, searchable digital formats using OCR and automatic classification.

Objective
 Digitize — run OCR across scanned or photographed documents, including noisy scans, handwriting, stamps,
and multiple languages/scripts.
 Classify — automatically sort documents into categories such as application forms, ID proofs, certificates, or
correspondence.
 Extract — pull key structured fields — names, dates, IDs, amounts — into a searchable database.
 Verify — flag low-confidence extractions for manual review instead of silently guessing.

Expected Outcomes
 A working OCR-plus-classification pipeline tested on a representative document set
 Structured, searchable records exported in a standard format (CSV/JSON)
 Accuracy benchmarks for both extraction and classification confidence
 LEGALTECH

06 The Record That Everyone Holds and Nobody Trusts

Reconciling Revenue, Registration & Survey land records

Land records in India are maintained separately by three departments — Revenue, Registration, and Survey — and
they can disagree on the same property: one record may name a different owner than another, and the survey map
may show a different boundary altogether.

Even with digitization and unique parcel IDs like ULPIN/Bhu-Aadhaar, there's still no common system to decide which
department's record is correct — fueling disputes, delays, and complications in loans and court cases.

What the Hackathon Wants

Build a system that compares conflicting land records, identifies inconsistencies, and provides a trusted resolution
process — while still following India's existing property laws.

IN ONE LINE
Create a trusted system that helps Revenue, Registration, and Survey departments agree on the correct
information about a property.
 EDTECH

07 The Tutorial That Taught Nothing

Turning watch-time into verified skill

Millions of students learn to code from online tutorials, but watching a video is not the same as understanding it. A
student can finish 200 hours of content and still be unable to build anything independently — because platforms
measure watch time and course completion, not actual comprehension.

What the Hackathon Wants

Build a system that tests whether a student can actually apply what they learned — before letting them move on.

 Works with existing videos, without changing them

 Gives students practical, contextual tasks or challenges
 Measures real understanding and skill, not just completion
 Provides verified proof of competency

IN ONE LINE
Build a system that converts “I watched the tutorial” into “I can actually do it.”
 PUBLIC SAFETY

08 The Gap Between the Train and the Cab

Continuous night-time safety across transport modes

Women travelling at night are usually monitored only inside one transport system — a train or a cab. The real risk
lives in the transitions: leaving a station, walking through an unsafe stretch, catching an auto, then walking home
from the drop-off point. Separate transport systems don't share safety data, and monitoring quietly drops at each
handoff.

What the Hackathon Wants

Build a continuous safety system that keeps a woman's safety session active across trains, cabs, autos, and walking
segments.

 Works on a normal smartphone

 Continues monitoring automatically through transport changes
 Never requires restarting the session at each transition
 Works even when railway, government, and private operators don't share data directly

IN ONE LINE
Create a safety system that protects a woman throughout her entire journey, not just inside one transport
system.
 GST & MSME TECH

09 The CA in Your Pocket That Does Not Exist

A smartphone GST assistant for small traders

Small businesses handle hundreds of invoices a month through WhatsApp, PDFs, and printed bills, and lean on a
Chartered Accountant to catch issues like a wrong HSN code or missed Input Tax Credit. But the GST portal itself
never explains what's wrong, why it matters, how much money is at stake, or what to do next.

What the Hackathon Wants

Build a smartphone-based GST assistant that ingests invoice and GSTR-2B data, detects mismatches, and explains the
fix in plain language.

 Handles messy invoice data across formats

 Identifies GST/ITC-related issues
 Explains problems in easy-to-understand language
 Tells the trader exactly what to do next
 Works without direct GST portal API access — dummy GSTR-2B data is fine for prototyping

IN ONE LINE
Build a “CA in your pocket” that helps small traders find GST/ITC problems and tells them how to fix them.
 PRIVACY & AI GOVERNANCE

10 The Right to Be Forgotten That Goes Nowhere

A unified privacy control center for AI platforms

People use many AI assistants, and each stores conversations and personal data separately. India's DPDP Act 2023
grants a right to deletion, but there's no simple way to see what's stored where — or to confirm a deletion actually
happened.

What the Hackathon Wants

Build a unified privacy assistant that helps users understand and manage the personal data they've shared with
different AI platforms.

 Shows what each AI assistant may have stored

 Explains consent and privacy terms in plain language
 Lets users create and manage deletion or correction requests
 Provides a record or proof of request status
 Works without direct API access to multiple AI platforms — mock/synthetic data is fine

IN ONE LINE
Build a “privacy control center” that helps users see, request deletion of, and track their personal data
across multiple AI assistants.
 CLIMATE & CITIES

11 Agri Smart — Smart Digital Farming Platform

Digital planning, monitoring & market access for farmers

Farmers often plan crops, monitor fields, manage resources, and sell produce using traditional methods — leading to
wasted water and fertilizer, lower productivity, and poor market decisions.

What the Hackathon Wants

A digital farming platform that helps farmers plan, monitor, manage resources, and connect with markets — turning
useful information into better decisions and higher productivity.

 Crop planning assistance using weather and soil inputs

 Farm monitoring dashboard with resource-usage alerts (water, fertilizer)
 Market price and access recommendations for selling produce

IN ONE LINE
Build a smart digital platform that helps farmers plan, monitor crops, use resources efficiently, and access
better markets.
 CLIMATE & CITIES

12 Crowdsourced Accessibility Mapping Platform for

Urban India
A Waze-style map for wheelchair & mobility-impaired users

Wheelchair users and people with mobility impairments lack reliable, up-to-date information on which public spaces
are actually accessible — ramps, restrooms, step-free metro entries, curb cuts.

Expected Outcome
 Map-based app (Google Maps / OpenStreetMap APIs) for pinning locations
 Accessibility tags, photos, and ratings attached to each pin
 Basic route planning that avoids inaccessible paths
 Upvote / flag system to keep community data trustworthy
 CLIMATE & CITIES

13 Ward-Level Urban Risk Mapper

Flood & accident hotspot prediction for municipal action

City wards rarely know where the next flood or accident is likely to happen until it does. Combine drainage, rainfall,
road condition, tree cover, and complaint data to predict flood-prone and accident-prone micro-zones at the ward
level, and suggest low-cost municipal interventions.

Objective
 Aggregate drainage, rainfall, road-condition, tree-cover, and citizen-complaint datasets at ward level
 Build a predictive model that flags flood-prone and accident-prone micro-zones
 Recommend low-cost, prioritized municipal interventions for each flagged zone

Expected Outcome
 A ward-level risk map highlighting flood and accident hotspots
 A ranked list of recommended interventions per zone
 An estimated before/after impact for each suggested intervention
 SPACE & DEEP TECH

14 Smart Predict — AI-Based Prediction System

Open track: pick your own real-world prediction problem

Many real-world problems involve large volumes of data and repetitive tasks that fixed-rule software can't handle.
This is a deliberately open track — build an AI/ML-based system that learns from data to automate tasks, predict
outcomes, personalize services, or support intelligent decision-making for a real-world problem of your team's
choosing.

Objective
 Identify a real-world problem well suited to prediction, automation, or personalization
 Source and prepare a relevant dataset for that problem
 Build and train an ML model appropriate to the chosen problem
 Demonstrate the model's outputs through a working interface or dashboard

Expected Outcome
 A working AI/ML prototype solving a clearly defined real-world problem
 Documented reasoning for the problem choice, model choice, and evaluation approach
 Basic accuracy or performance metrics for the chosen use case
 GOVTECH

15 Medicine Shortage & Price Monitoring System

Early warning for drug unavailability and price gouging

Essential medicines like insulin and thyroid drugs sometimes vanish from pharmacies without warning — driving up
prices and encouraging illegal stockpiling or resale. No single system currently flags where shortages or unusual price
spikes are happening.

What the Hackathon Wants

 Pharmacies update stock levels; patients report shortages or high prices
 System analyzes reports to identify affected areas and shows them on a map
 Automated alerts to government authorities, NGOs, and medicine suppliers

IN ONE LINE
Track medicine availability and prices, detect shortages early, map affected areas, and alert the right
people before it becomes serious.
 GOVTECH

16 AYUSH Startup Registration Platform

One portal for the entire AYUSH registration journey

AYUSH-sector startups face a slow, complicated registration process spread across disconnected systems, with no
single place to submit applications, upload documents, check requirements, or track status.

Platform Features
 Single-window application submission with document upload
 Automated compliance and requirements checklist per registration type
 Real-time status tracking for applicants
 In-platform communication channel with registration officials

IN ONE LINE
Build a single online platform where AYUSH startups can register, submit documents, check compliance,
track applications, and get updates until registration is complete.
 GOVTECH

17 Citizen Grievance Triage Engine

Cross-department complaint routing and true-resolution tracking

Citizens struggle to know where to complain, whether a complaint was received, or when it will be resolved. Build an
engine that reads complaints from multiple government portals, detects duplicate or related complaints across
departments, assigns the correct agency, and tracks whether the issue was actually resolved — not merely closed.

Objective
 Ingest complaints from multiple government portals into one system
 Detect duplicate or related complaints filed across departments
 Automatically route each complaint to the correct department or agency
 Track status through to verified resolution, not just closure

Expected Outcome
 A unified complaint dashboard spanning multiple departments
 De-duplication and correct-routing accuracy metrics
 A resolution-tracking view that distinguishes “closed” from “actually resolved”
 FINTECH & SOCIAL IMPACT

18 Real-Time Sign Language Translator for Local

Commerce
ISL-to-speech translation for everyday transactions

India's Deaf community routinely faces communication barriers in everyday transactions — buying groceries,
bargaining at markets, asking for directions. Build a mobile app that uses the phone's camera to translate Indian Sign
Language gestures into speech or text in real time, and converts spoken input back into text or simplified sign cues.

Expected Outcome
 Hand-landmark detection (e.g. MediaPipe) feeding a gesture classifier
 Simple TTS/STT layer for the reverse (speech-to-sign) direction
 Bonus: offline support and regional sign-variation handling
 FINTECH & SOCIAL IMPACT

19 Verified Local-Connect Platform for Cultural

Tourism
Matching travelers with background-verified local hosts

Solo and first-time travelers often struggle to access authentic local experiences and trustworthy on-ground guidance.
Build a platform that matches tourists with background-verified local volunteers or hosts for short, informal meetups
— conversation, local tips, informal walking guidance.

Expected Outcome
 Host verification/rating flow (ID upload + manual approval works for a demo)
 In-app scheduling between traveler and host
 Safety feature: live location sharing or an SOS button
 FINTECH & SOCIAL IMPACT

20 Digital Marketplace for Ethical Waste Aggregation

Fair-price, direct-access marketplace for waste-pickers

Informal waste-pickers power India's recycling economy but are routinely underpaid by middlemen due to a lack of
price transparency and market access. Build a marketplace connecting waste-pickers and aggregators directly with
recycling units and scrap buyers.

Expected Outcome
 Reference material pricing, pickup/scheduling flow, transaction history
 Voice input, icons over text, and regional-language support for low digital literacy
 Simple listing plus chat/call-based negotiation is enough for a demo
 FINTECH & SOCIAL IMPACT

21 Bank-Loan Due-Diligence Assistant for Rural

Property Loans
Title-risk screening before rural & semi-urban disbursement

Rural and semi-urban property loans carry hidden title risk that's easy to miss during manual review — mismatched
names, unrecorded mutations, or a survey boundary that doesn't match the tax record. Build an assistant that checks
whether land-title records, tax receipts, mutation entries, and survey maps align before a loan is disbursed, and
highlights hidden title-risk indicators.

Objective
 Cross-check land-title records, tax receipts, mutation entries, and survey maps for a given property
 Flag mismatches or inconsistencies that indicate title risk
 Generate a simple due-diligence report loan officers can act on before disbursement

Expected Outcome
 An automated title-risk report for a sample set of rural/semi-urban properties
 Clear flags for missing, mismatched, or high-risk records
 Reduced manual verification time for loan officers
 LEGALTECH

22 Case Document Consistency Checker

Contradiction & gap detection for courts and legal aid

A single case file can run to dozens of documents — FIRs, witness statements, charge sheets, annexures — and small
inconsistencies between them (a misspelled name, a missing date, a contradicted timeline) can stall a case or weaken
it in court. Design a checker for courts and legal aid offices that compares these documents, identifies contradictions
or missing fields, and generates a risk report for clerical and procedural errors.

Objective
 Parse FIRs, witness statements, charge sheets, and annexures for a given case
 Cross-check facts, names, dates, and cited sections across documents
 Flag contradictions, missing fields, or procedural errors
 Generate a structured risk report for court or legal-aid review

Expected Outcome
 A working prototype that ingests a sample case-file set and flags inconsistencies
 A clear, prioritized risk report separating clerical from substantive issues
 Reduced manual cross-referencing effort for legal aid staff
 EDTECH

23 Parental Screen-Time Monitoring & Digital

Wellbeing Dashboard
Healthy-habit nudges over blunt restriction

Parents often lack visibility and actionable tools to manage children's screen time in a way that builds healthy habits
rather than just restricting access. Build a paired parent-child companion app that tracks screen time and app-
category usage, letting parents set flexible schedules and limits.

Expected Outcome
 Android: UsageStatsManager / Accessibility APIs on the child device, synced to a parent dashboard
 At least one nudge mechanism — reminders, earned break extensions, or gamified goals — instead of a pure
lockout
 EDTECH

24 Student Learning-Gap Resolver

Diagnosing prerequisite gaps, not just tracking scores

A student who keeps failing algebra questions may not have an algebra problem at all — the real gap might be a
missed prerequisite from two chapters back. Map syllabus topics to assessment results, detect where students
repeatedly fail on prerequisite concepts, and generate personalized micro-remediation paths for teachers and
learners.

Objective
 Map syllabus topics to their prerequisite concepts and related assessment questions
 Analyze student assessment results to detect repeated-failure patterns
 Identify the specific prerequisite gap behind each recurring mistake
 Generate a personalized micro-remediation path per student

Expected Outcome
 A gap-analysis view for teachers showing where a class is struggling
 Personalized remediation paths for individual students
 A before/after comparison showing improvement after remediation
 PUBLIC SAFETY

25 Multi-Source Incident Verification System

Separating genuine emergencies from noise, in real time

Emergency responders are flooded with reports from disconnected sources — police calls, CCTV feeds, social media,
citizen alerts — with no easy way to tell which ones describe the same incident, which are false alarms, and which
genuinely need an urgent response. Build a system that merges police calls, CCTV metadata, social-media alerts, and
emergency reports to identify likely false alarms, duplicate incident reports, and genuine high-priority emergencies in
real time.

Objective
 Ingest police calls, CCTV metadata, social-media alerts, and emergency reports
 Correlate reports that refer to the same underlying incident across sources
 Score each incident for likely authenticity and priority
 Surface genuine high-priority emergencies to responders in real time

Expected Outcome
 A working prototype that merges at least two to three of the listed sources
 De-duplication and false-alarm filtering accuracy metrics
 A real-time, prioritized incident feed for a demo dispatcher view
 PRIVACY & AI GOVERNANCE

26 Privacy-Safe AI Audit Layer for Public-Sector

Services
Catching data over-collection before it happens

Public-sector digital services routinely ask for more citizen data than the service actually needs, often without anyone
flagging it. Develop an audit layer for public-sector digital services that detects when citizen data is being over-
collected, explains which fields are unnecessary for the service, and generates compliance warnings based on purpose
limitation and data minimization.

Objective
 Scan the data fields collected by a public-sector digital service
 Compare each field against the service's stated purpose
 Flag fields that appear unnecessary or excessive for that purpose
 Generate a plain-language compliance warning for reviewers

Expected Outcome
 A working audit tool tested against at least one sample public-service data flow
 A clear compliance report flagging over-collected fields
 Recommendations aligned with data-minimization principles