Final year project ideas
for computer science students

Difficulty: ★ manageable for most teams, ★★ moderate, ★★★ ambitious.

1. Indian-language document assistant (RAG)

Ask questions of government circulars or textbooks in Hindi or a regional language; answers cite the source page. ★★★ — retrieval, embeddings, evaluation of answer accuracy.

2. Crop disease detector

Classify leaf images and suggest treatment, working on low-end phones. ★★ — image classification, model compression. Public plant-disease datasets are available on Kaggle².

3. UPI-style transaction fraud detector

Detect anomalous transactions in a synthetic or public dataset, with explanations. ★★ — imbalanced data, precision–recall trade-offs.

4. Air quality forecaster

Forecast city AQI from historical readings and weather, using open data from data.gov.in¹. ★★ — time series.

5. Resume and job-description matcher

Score fit and suggest missing skills, with attention to bias. ★★ — NLP, ranking, fairness discussion.

6. Sign language to text

Recognise a set of Indian Sign Language gestures from video. ★★★ — pose estimation, sequence models.

7. Fake news and claim checker

Match claims to trusted sources and show evidence. ★★★ — retrieval, classification, careful evaluation.

8. Handwritten answer sheet evaluator

OCR handwritten answers and suggest marks against a rubric for a teacher to review. ★★★ — OCR, NLP, human-in-the-loop design.

9. Accessible campus navigation app

Step-free routes, lifts and ramps across your campus. ★★ — maps, graph search, accessibility testing.

10. Privacy-first personal health record

Store reports encrypted on device, share selectively with a doctor by link that expires. ★★ — encryption, consent design.

11. Blood and organ donor matching

Match requests to nearby compatible donors with verified profiles. ★★ — geo-queries, notifications, privacy.

12. Smart attendance with face verification

Classroom attendance with liveness checks and consent. ★★ — computer vision, privacy discussion.

13. Offline-first field data collection

Survey app that works offline and syncs conflicts safely. ★★★ — sync, conflict resolution.

14. Real-time collaborative whiteboard

Multiple users draw together live. ★★★ — WebSockets, CRDTs or operational transforms.

15. Local marketplace with recommendations

Buy and sell within a campus or area, with a recommendation engine. ★★ — full stack, collaborative filtering.

16. Event ticketing with QR check-in

Tickets, payments in test mode and offline QR scanning at the gate. ★ — full stack, QR, payments integration.

17. Phishing URL classifier

Classify URLs and pages as phishing with a browser extension. ★★ — feature engineering, ML, extension development.

18. Password strength and breach checker

Check strength and known-breach exposure using k-anonymity hashing so passwords never leave the device. ★ — hashing, privacy-preserving design.

19. Network intrusion detection

Detect attacks in network traffic datasets with ML and rules. ★★★ — networking, ML.

20. Secure file sharing with expiry

End-to-end encrypted sharing with expiring links and audit logs. ★★ — cryptography, web security against the OWASP Top 10⁵.

21. Smart irrigation with soil sensors

Sensors decide when to water; dashboard and alerts on phone. ★★ — microcontrollers, MQTT.

22. Parking space detection

Camera-based free-space detection for a campus car park. ★★ — computer vision.

23. Home energy monitor

Track appliance-level energy use and suggest savings. ★★ — sensors, time series.

24. Elderly fall detection

Phone or wearable sensors detect falls and alert family. ★★ — signal processing, ML.

25. Public transport crowd estimator

Estimate bus or station crowding from anonymised Wi-Fi or camera counts. ★★★ — privacy-preserving sensing.

26. Mini search engine

Crawl a site, build an inverted index and rank results. ★★ — indexing, ranking.

27. Distributed key-value store

Replication and leader election with failure testing. ★★★ — distributed systems.

28. Compiler for a tiny language

Lexer, parser, type checker and code generation for a small language. ★★★ — compilers.

29. Load balancer and rate limiter

Build and benchmark algorithms under load. ★★ — networking, performance testing.

30. Code plagiarism detector

Compare submissions using token and AST similarity. ★★ — algorithms, NLP on code.

• README first — what it does, why, how to run it, screenshots and a demo video link³
• Architecture diagram — one picture of components and data flow
• Clear structure — separate folders for app, models, tests, docs
• Tests — at least for the core logic, with instructions to run them
• Reproducible results — scripts and fixed random seeds for ML experiments
• Licence — pick one so others know how they can use it⁴
• No secrets in the repo — keys go in environment variables
• Honest limitations — a section on what doesn't work yet

1. Weeks 1–3: choose the problem, talk to users, survey existing work, pick data.
2. Weeks 4–5: design architecture, data model and evaluation plan.
3. Weeks 6–12: build in two-week iterations; demo to your guide each time.
4. Weeks 13–14: evaluate with real users or held-out data; fix the top issues.
5. Weeks 15–16: documentation, demo video, report and viva rehearsal.

Want to build for phones specifically? See mobile app ideas for a final year project. Need something you can build in a weekend? See hackathon project ideas.

RAG — retrieval-augmented generation, answering from retrieved documents. Baseline — a simple model to compare against. CRDT — a data structure for conflict-free collaborative editing. Inverted index — a map from words to documents containing them. README — the front page of a code repository.