🏆 Problem Statements Worth Building

Real-world problems for hackathons, college projects, and developer portfolios. Ready to build. Open for contributions.

This is the largest curated collection of real-world problem statements (46 problems across 5 tracks) — built for hackathons, college major/minor projects, capstones, and developer portfolios. Every problem is grounded in actual data, peer-reviewed research, and serves real stakeholders — not hypotheticals.

---

Stop building chatbots and CRUD dashboards. Build something that actually matters.

The 5 Tracks

Track	Format	Build Time	Best For
🌍 Global South Impact	10 AI/ML problems	5–18 months	Teams with ML + domain expertise
🇺🇸 US Civic Tech	10 LLM/data problems	6–16 weeks	Teams building for consumer impact
🇮🇳 India Impact	5 AI/agent problems	8–16 weeks	Solo devs with DPI API experience
🧠 Frontier AI Platforms	10 AI-governance/health problems	2–6 months	Solo devs tackling systemic risks
⚡ Rapid Prototypes	11 engineering problems	2–6 weeks	Solo devs, weekend hackathons

---

The Problem With Hackathons

Most hackathon prompts fall into three failure modes:

Failure Mode	Example	Why It Fails
🫗 Underspecified	"Build something with AI for healthcare"	Too vague to start building. Teams spend 50% of the event deciding what to do.
🏗️ Over-engineered	"Build a federated learning platform for genomics"	Requires months of prior infrastructure. Unbuildable in a weekend.
🎭 Fake	"Airbnb for pets"	Doesn't solve a real problem. Judges can smell it. Won't survive outside the classroom.

This repo fixes all three. Every problem statement here is:

• ✅ Real — Grounded in actual data sources, government datasets, published research, and quantified stakeholders
• ✅ Concrete — Specific enough to start building in one sitting. Datasets linked. Papers cited. Adjacent OSS mapped.
• ✅ Impactful — Affects millions of people or billions of dollars. Every problem has a built-in "why this matters" case.
• ✅ Scoped — Clear MVP timeline with success criteria. You know exactly what "done" looks like.
• ✅ Feasible — Build-time estimated by complexity. AI vs non-AI labeled. Solo-friendly options available.

---

Who This Is For

Audience	Why This Repo Exists For You
Hackathon participants	Stop wasting 4 hours debating ideas. Pick a battle-tested problem from the list and start building in 15 minutes.
Hackathon organizers	Ditch the vague theme. Drop real problem statements that produce working, fundable prototypes — not another todo app.
College students (minor projects)	Need a 1-semester project that actually works? Every Rapid Prototype and most US Civic Tech problems are scoped to build in 6–16 weeks with clear success criteria. No fluff.
College students (major projects / capstone)	Looking for something with enough depth for a year-long thesis? Global South Impact and Frontier AI Platform problems have research citations, data sources, and evaluation frameworks built in.
Portfolio builders	Want a GitHub that gets you hired? These problems produce demonstrable, real-world projects — not another CRUD dashboard. Recruiters pay attention when you can say "I built a fraud detection system used by X people."
University clubs & CS departments	Give students problems that connect coursework to real-world impact. Every problem links to papers and datasets with clear scope for a semester project.
Open-source contributors	Find a problem you care about and start building. The repo is designed for low-friction contribution.
Indie developers & founders	Each problem statement is a potential startup idea with identified stakeholders, market gaps, and monetization paths.
NGOs & government agencies	Submit problems from the field. The _PROBLEM_TEMPLATE.md makes this a 15-minute contribution.

---

🎓 For College Students & Portfolio Builders

This repo was built for you as much as for hackathon teams. Here's how to pick the right problem for your context.

By Academic Scope

Project Type	Timeframe	Recommended Track	Why
Minor Project (1 semester)	8–16 weeks	Rapid Prototypes or US Civic Tech	Scoped for solo development. Clear MVP. No research phase needed.
Major Project (2 semesters)	16–32 weeks	India Impact or Frontier AI Platforms	Real API integration, multiple components. Enough depth for a proper evaluation.
Capstone / Thesis (1 academic year)	6–18 months	Global South Impact or Frontier AI Platforms	Research-grounded. Papers to cite. Datasets to analyze. Genuine open-endedness.
Portfolio project	As long as it takes	Any track — pick what excites you	Pick something you can demo end-to-end. A solved problem with real data beats a half-baked complex idea.

Choosing the Right Problem — A Framework

Not every problem fits every student. Use this filter:

Academic level → Available time → Skills you want to learn → Domain you care about → Pick 3 problems → Read all 3 → Build the one that excites you most

Minor project checklist:

• Can I build the MVP in <10 weeks?
• Do I already know 70% of the tech stack?
• Is there a working demo I can show in 2 minutes?
• Does it produce something someone could actually use?

Major project / thesis checklist:

• Are there enough research papers to build a literature review? (→ Global South, Frontier AI)
• Is there real data I can access? (→ all problems link datasets)
• Can I frame an evaluation (A/B test, benchmark, user study)?
• Will I still be excited about this in month 5?

Portfolio Strategy

A GitHub portfolio with one of these problems says more than five tutorial projects. Here's why:

Instead of building…	Build one of these…	Why it wins
A weather dashboard (tutorial clone)	CivicFeed — Public Comment AI	Production NLP + government API. Real stakeholder.
A Twitter clone (everyone's done it)	FOIAbot — FOIA Assistant	Legal tech. Document intelligence. Zero competitors.
A CRUD todo app	Village Grain Bank — WhatsApp Banking for Farmers	Offline-first architecture. Real users. Actual impact.
A chatbot wrapper	DecodeMyBill — Medical Bill Parser	Domain expertise signal. Healthcare + OCR + LLMs.
An e-commerce site	Vyavastha — MSME Compliance Copilot	B2B SaaS potential. Multiple API integrations.

Pro tip: Contribute your implementation notes back as a PR. A repo that shows both "built this" and "documented how" is the kind of candidate teams fight over.

---

The Problem Sets

🌍 Global South Impact — AI/ML for the Developing World

10 problems that solve invisible infrastructure challenges in low-resource settings — where commercial SaaS doesn't compete because customers can't pay market rates.

#	Problem	Impact	Build Time	Tags
01	Maternal Health Risk Stratification	287K maternal deaths/year	6–8 mo	health ML offline
02	Public Procurement Fraud Detection	$1.3–4T lost/year	5–7 mo	governance GNN
03	Informal Waste Sector Platform	15–20M informal workers	6–8 mo	environment CV
04	Post-Harvest Loss Intelligence	30–40% food lost	7–9 mo	agriculture IoT
05	Harmful Algal Bloom Early Warning	60% US lakes at risk	8–10 mo	water remote-sensing
06	Scientific Reproducibility Engine	$28B/year wasted	8–12 mo	science LLM
07	Offline Crop Disease Diagnostics	500M+ farmers	5–7 mo	agriculture TFLite offline
08	Groundwater Depletion Forecasting	2B+ people	12–18 mo	water climate satellite
09	School Resource Allocation Optimizer	65M US students	5–7 mo	education optimization
10	Climate-Resilient Housing Design	1B+ in slums	10–14 mo	housing climate gen-AI

🏁 Fastest MVP: Offline Crop Disease Diagnostics (TFLite on-device, 5–7 months, solo-buildable)

---

🇺🇸 US Civic Tech — Consumer Advocates for Broken Systems

10 problems that navigate America's most opaque consumer-facing systems — medical billing, workers' comp, family court, FOIA, public comments, predatory lending, housing assistance, and school funding. Every one affects millions, has public data available, and currently has no open-source consumer advocate.

#	Problem	Novelty	MVP Time	Why It Exists
01	CivicFeed — Public Comment Intelligence	9/10	6–8 wks	Agencies spend 3–6 person-weeks per docket reading comments
02	Workers Compass — Claim Navigator	10/10 🏆	8–10 wks	$50B industry, zero consumer competitors
03	FOIAbot — Public Records Assistant	9/10	6 wks	Journalists spend months per FOIA request
04	DecodeMyBill — Medical Bill Intelligence	7/10	10–12 wks	80% of medical bills have errors
05	ProSe Navigator — Family Court Assistant	8/10	8–12 wks	70–80% self-represented in family court
06	PredatoryGuard — Financial Analyzer	9/10	6–8 wks	$10B+ in scam losses/year
07	UtilityCoach — Energy Assistance Navigator	8/10	6 wks	80% of eligible households never apply for LIHEAP
08	HousingKey — Housing Program Navigator	8/10	10–12 wks	20M+ households with worst-case housing needs
09	InformedYou — Consent Simplifier	7/10	6 wks	Consent forms written at college level; avg US adult reads at 8th grade
10	SchoolEquityWatch — Funding Transparency	7/10	12–16 wks	High-poverty districts receive 16% less funding — data is buried

🏁 Most defensible: Workers Compass ($50B workers' comp industry, zero consumer-facing competitors)

---

🇮🇳 India Impact — AI on India's DPI Layer

5 AI problems targeting India's Digital Public Infrastructure (DPI). Unlike US civic tech (fragmented) or generic Global South problems (pre-DPI), India's plumbing is built — APIs exist for land records, court cases, agricultural prices, and 740+ government schemes. But the intelligent application layer is missing.

#	Problem	Type	MVP	Startup Potential
01	Kisaan Marg — Mandi Price Intelligence	Agmarknet API + LLM agent + WhatsApp	10–12 wks	₹10,000 Cr/yr leakage
02	Vyavastha — MSME Compliance Copilot	Regtech agent + API orchestration	12–16 wks	6.45Cr customers
03	JalGuru — Water Quality Intelligence	Geospatial ML + alerts	8–10 wks	Public health mission
04	Nyaya Sahayak — Court Case Navigator	eCourt API + LLM + WhatsApp	8–12 wks	52M pending cases
05	Sarthak — Government Scheme Agent	Eligibility engine + DigiLocker	8–10 wks	₹7.67L Cr in schemes

🏁 Best startup bet: Vyavastha (MSMEs pay ₹13–17L/yr for compliance — will pay for a cheaper alternative)

---

🧠 Frontier AI Platforms — AI Governance, Health & Systemic Risk

10 problems at the frontier of AI application — algorithmic auditing, antimicrobial resistance, clinical trial equity, dementia care, wildfire resilience, and more. These sit where AI capability meets systemic risk. Each problem has a clear regulatory or scientific framework behind it.

#	Problem	Domain	Build Time	Stack
01	Algorithmic Bias Auditing Platform	AI governance	2–3 mo	AIF360, FairLearn, SHAP
02	Youth Mental Health Crisis Triage	Mental health	3–4 mo	NLP, risk models
03	Clinical Trial Matching & Patient Equity	Health equity	3–4 mo	LLM, OMOP CDM, FHIR
04	Homelessness Prevention Early Warning	Social services	4–5 mo	ML, court data
05	AMR Surveillance & Prescribing Support	Global health	4–6 mo	WHO data, CLSI, FHIR
06	Dementia Caregiver Decision Support	Aging	3–4 mo	RAG, clinical guidelines
07	Food Waste Surplus Redistribution	Climate	3–4 mo	OR, supply-chain
08	Wildfire Risk & Community Preparedness	Climate	3–5 mo	Geospatial ML, satellite
09	Perinatal Mental Health Platform	Maternal health	2–3 mo	NLP, screening tools
10	SMB Cybersecurity Compliance	Cybersecurity	2–3 mo	NIST, CMMC, LLM

🏁 Fastest build: Algorithmic Bias Auditing or Perinatal Mental Health (2–3 months with existing frameworks)

---

⚡ Rapid Prototypes — Build for Impact in a Weekend

11 non-AI ideas scoped for solo developers and weekend hackathons. No AI, no computer vision — just solid engineering that changes lives. Pure CRUD, maps, WhatsApp bots, and data pipelines.

#	Problem	Stack	Build Time	Learning Curve
01	Village Grain Bank Manager	Twilio + CRUD + Inventory	2–3 wks	★☆☆☆☆
02	Medicine Stock Visibility	WhatsApp/SMS + Inventory	2–4 wks	★☆☆☆☆
03	Infrastructure Defect Reporter	Maps + Escalation Workflow	3–4 wks	★★☆☆☆
04	Procurement Data Quality Monitor	Data Pipeline + Dashboard	3–4 wks	★★☆☆☆
05	Informal Worker Skills Passport	Offline Mobile + QR	4–6 wks	★★★☆☆
06	School Resource Transparency Map	Offline Forms + Maps	4–6 wks	★★★☆☆
07	Annapurna — PDS Tracker	WhatsApp + Maps + API	4–6 wks	★★★☆☆
08	RathLink — Waste Worker Platform	QR + Offline Mobile	4–6 wks	★★★☆☆
09	BhuLekh — Land Records App	API Search + Maps	3–4 wks	★★☆☆☆
10	Setu — Government Form Assistant	Form Engine + DigiLocker	3–4 wks	★★☆☆☆
11	JalSathi — Water Testing Network	Maps + SMS + Crowdsource	4–6 wks	★★★☆☆

🏁 24-hour hackathon pick: Village Grain Bank Manager or Medicine Stock Visibility

---

How to Use This Repo

🎯 For Hackathon Teams (5-minute start)

1. How much time do you have?
   • 24–48 hours → Go straight to Rapid Prototypes
   • 1–4 weeks → Pick a US Civic Tech problem
   • 1–6 months → Your pick — all tracks open
2. What's your team's skill stack?
   • Computer vision → Global South Impact #3, #7
   • LLMs / NLP → US Civic Tech #1–#10 (all leverage LLMs differently)
   • Full-stack, no ML → Rapid Prototypes (pure engineering impact)
3. Read the problem statement → datasets linked, papers cited, OSS to build on
4. Ship the MVP → every problem has clear success criteria
5. Contribute back → PR your implementation notes

🏫 For Organizers & Educators

Pick problems proportional to your event or semester duration:

Event Length	Recommended Track	Rationale
24–48 hours	Rapid Prototypes	Fully buildable solo over a weekend
1 week	US Civic Tech (select 3–5)	LLM-powered, fast iteration
1–3 months	Global South Impact	Real ML, CV, or systems project
Semester-long	Any + mentorship	Research-grade implementation

🎓 For College Students

Quick pick by your situation:

• Minor project this semester → Rapid Prototypes — 2–6 week builds, no ML required, solo-scoped. Village Grain Bank or Medicine Stock Visibility are perfect starter projects.
• Major project (year-long) → US Civic Tech or India Impact — 8–16 week scope leaves room for literature review, implementation, and evaluation. Workers Compass or Vyavastha have startup-level depth.
• Capstone / thesis → Global South Impact or Frontier AI Platforms — 5–18 month scope, research-grounded, publishable results. Pick one with datasets you can access and papers you can cite.
• Portfolio piece to get hired → Any track, but prioritize problems that let you show: architecture decisions, testing strategy, CI/CD, and a live or recorded demo.

Academic workflow:

1. Browse by domain → INDEX.md lists all problems by domain, skill, and geography
2. Read 3 problem statements → find one that excites you
3. Check data availability → every problem links datasets upfront
4. Draft a project proposal → the problem statement IS your proposal (problem, data, methods, success criteria)
5. Build the MVP → success criteria tell you exactly what "done" looks like
6. Write the report → problem statement gives you the introduction and related work
7. Submit feedback → PR your experience to help the next student

🤝 For Contributors

See CONTRIBUTING.md. Three ways to contribute:

1. Submit a new problem — Use _PROBLEM_TEMPLATE.md (15-minute task)
2. Improve an existing one — Add datasets, fix errors, add adjacent OSS
3. Implementation notes — Built a solution? Share architecture, pitfalls, and lessons learned

---

What Makes This Different From Every Other Hackathon Problem List

Feature	This Repo	Typical Hackathon Problems
Real data sources	✅ Every problem has linked datasets and API endpoints	❌ Vague or absent
Academic grounding	✅ 200+ peer-reviewed papers cited across 46 problems	❌ None
Stakeholder analysis	✅ Quantified: who's affected, at what scale, with sources	❌ None
MVP scope	✅ Build time estimated in weeks or months per problem	❌ "Build something cool"
Open-source adjacencies	✅ 80+ existing OSS projects mapped per track	❌ No landscape awareness
AI vs non-AI labeled	✅ Clear labeling so teams pick by available skills	❌ No guidance
Novelty scoring	✅ Every US Civic Tech problem scored 7–10/10	❌ None
Monetization paths	✅ B2G / B2B / open-core models for each prototype	❌ None
Failure mode analysis	✅ "Why existing solutions fail" for every problem	❌ None
Success criteria	✅ Checklist-style "what done looks like" per problem	❌ None

---

Quick Navigation by Build Time

Build Time	What You Can Build
2–3 weeks	Village Grain Bank, Medicine Stock Visibility
3–4 weeks	Infrastructure Defect Reporter, Procurement Data Quality Monitor, BhuLekh, Setu
4–6 weeks	Informal Worker Skills Passport, School Resource Transparency Map, Annapurna, RathLink, JalSathi
6–8 weeks	CivicFeed, FOIAbot, PredatoryGuard, UtilityCoach, InformedYou
8–12 weeks	Workers Compass, DecodeMyBill, ProSe Navigator, HousingKey, Kisaan Marg, JalGuru, Nyaya Sahayak, Sarthak
12–16 weeks	SchoolEquityWatch, Climate-Resilient Housing, Vyavastha
5–7 months	Procurement Fraud, Offline Crop Disease, School Resource Optimizer
6–8 months	Maternal Health, Informal Waste, Post-Harvest Loss
8–10 months	Harmful Algal Bloom Early Warning
8–12 months	Scientific Reproducibility Engine
12–18 months	Groundwater Depletion Forecasting

---

Selection Criteria

Every problem in this repo had to pass all of these filters:

• ❌ Not a chatbot wrapper, generic RAG, CRUD dashboard, note-taking tool, meeting summarizer, productivity app, AI code assistant, or social network clone
• ✅ Strong societal or economic impact (millions of people or billions of dollars)
• ✅ Existing datasets and/or published research literature to ground the problem
• ✅ AI/tech genuinely adds value — not a gimmick
• ✅ No saturated open-source solution already dominates the space
• ✅ Technically feasible for a motivated team (any skill level)

---

Repo Structure

hackathon-problem-statements/
├── global-south-impact/       # 10 AI/ML problems for the developing world
│   ├── README.md              # Track overview with quick-start guides
│   └── 10 problem statements  # Each: problem, data, papers, OSS, criteria
├── india-impact/              # 5 AI problems on India's DPI layer
│   ├── README.md              # Track overview with quick-start guides
│   └── 5 problem statements
├── us-civic-tech/             # 10 consumer/civic problems for the US
│   ├── README.md              # Track overview with novelty scores
│   └── 10 problem statements
├── rapid-prototype/           # 11 engineering problems, 2–6 weeks each
│   ├── README.md              # Track overview with week-by-week timelines
│   └── 11 problem statements
├── frontier-platforms/        # 10 AI-governance/health problems
│   └── 10 problem statements  # Each: problem, regulatory framework, data
├── _PROBLEM_TEMPLATE.md       # Template for submitting new problems
├── INDEX.md                   # Master index — filterable by time, skill, domain
├── CONTRIBUTING.md            # How to contribute
├── CODE_OF_CONDUCT.md         # Community standards
└── LICENSE                    # MIT — free to use, fork, build

---

Stats

Metric	Count
Total problem statements	46 (and growing)
Tracks	5 (Global South AI, US Civic Tech, India Impact, Frontier AI Platforms, Rapid Prototypes)
AI/ML problems	35
Pure engineering problems	11
Research papers cited	200+
Datasets and APIs linked	100+
Open-source adjacencies mapped	80+
Domains covered	20+ (health, agriculture, governance, education, water, housing, energy, labor, finance, science, legal, climate, transparency, infrastructure, environment, cybersecurity, mental health, food security, AI governance, clinical research)
Shortest build time	2 weeks
Longest build time	18 months

---

Frequently Asked Questions

Q: Are these actually buildable by a hackathon team?
Yes. Every problem has a clear MVP scope. The Rapid Prototypes are built specifically for 24–48 hour events. The US Civic Tech problems are designed for LLM-powered fast iteration over weeks. The Global South Impact problems require more time but have the largest potential impact.

Q: I don't know ML. Can I still contribute?
Absolutely. The entire Rapid Prototypes track requires zero ML — just solid full-stack engineering. Plus, many US Civic Tech problems can be tackled by integrating LLM APIs (you consume the model, you don't train one).

Q: Where do the problems come from?
Systematic landscape analysis across 30+ domains, 70+ papers, and 50+ candidate problems by AshayK003. Each problem was validated against real-world data sources, existing research, and stakeholder analysis.

Q: Can I submit a new problem?
Yes — and we want you to. Use _PROBLEM_TEMPLATE.md and open a PR. See CONTRIBUTING.md. We especially welcome problems from NGOs, government agencies, and domain experts.

Q: Can I fork this and use it for my own hackathon event?
Yes — it's MIT licensed. Use it, remix it, customize it for your event. We'd appreciate a shoutout but legally you don't need one.

Q: Has anyone built a solution for any of these?
Not yet — that's the point. These are open problems waiting for a team to step up. If you build one, submit implementation notes via PR.

Q: I'm a college student — can I use this for my minor/major project?
That's exactly what the 🎓 For College Students section is for. Every problem has clear scope, success criteria, datasets, and research citations — so you can spend your time building, not defining. The problem statement doubles as your project proposal introduction and related work section.

Q: Will putting one of these in my portfolio help me get a job?
A real-world project with actual data, a working demo, and documented architecture beats five tutorial clones every time. Recruiters notice when you can say "I built a tool that addresses an X-billion-dollar problem." The Portfolio Strategy section has specific recommendations.

---

License

MIT — use, fork, remix, build a company on it. If you start a business based on one of these, let us know — we'd love to feature it.

---

Developer Support

If these problem statements help you build something meaningful — or you just want to say thanks — consider supporting the developer:

@darkcharon3301

---

⭐ Star this repo — help the next team find a problem worth building.

View Full Index · Submit a Problem · Contribution Guide