About the Hackathon

Gemma for Bharat is an open AI hackathon hosted by GDG on Campus JIS University, co-hosted by GDG on Campus NIT, GDG on Campus GNIT, and GDG on Campus UEM, with ML Kolkata as our judging partner and OWASP JIS University as our swag partner.

The most useful AI in India will not come from copying what already exists. It will come from builders who understand a specific problem and ship a small, sharp tool that solves it. Gemma, Google DeepMind's family of open-weight models, gives you the raw capability. What you do with it is up to you.

Bring a real problem. Build a real solution. You have 9 days.

What You Will Build

Gemma for Bharat runs as a single track: Build with Gemma.

Your submission must be one of these four project types, each designed to run end-to-end inside a Kaggle Notebook. These are starting points, not straitjackets. Improvise, remix, and push further than what is described here. Judges will reward ambition that lands, not literal adherence to the brief.

1. Scripture and Philosophy Engine

Build a retrieval-augmented Q&A or study system over India's religious and philosophical texts. Work with translations and commentaries from the Bhagavad Gita, Upanishads, Ramayana, Mahabharata, Quran, Bible, Guru Granth Sahib, Tripitaka, Jain Agamas, Thirukkural, or any combination. Handle natural-language queries in English or an Indian language and return grounded answers with citations back to the source verse.

Push further: compare concepts across traditions, generate verse-by-verse commentary in modern language, build a thematic index of ideas like dharma, tawheed, karuna, seva, or ahimsa.

2. Indian Language and Script Bridge

Build a system that moves meaning across India's 22 official languages, classical languages (Sanskrit, Pali, Tamil, Prakrit), or historical scripts. Fine-tune or prompt Gemma to translate, transliterate, or bridge dialects. Examples: Sanskrit to modern Hindi with commentary, Perso-Arabic Urdu to Devanagari, code-mixed Hinglish or Tanglish normalization, Braj Bhasha or Awadhi to modern language, script conversion between Devanagari, Bengali, Tamil, Telugu, Kannada, Malayalam, Gurmukhi, Odia.

Push further: handle idioms and metaphors, preserve poetic meter, or build a real-time chat translator between two Indian language speakers.

3. Indian Sign Language and Accessibility Vision

Use Gemma's vision-capable variants to build a computer vision tool for accessibility, with a focus on Indian contexts. Examples: Indian Sign Language (ISL) gesture recognition and translation to text or speech, reading handwritten Devanagari or regional-script prescriptions, extracting structured data from Indian government forms (PAN, Aadhaar, ration card templates), describing culturally-familiar scenes for blind users, OCR on scanned palm-leaf manuscripts or old Indian-language documents.

Push further: real-time ISL translation from webcam, multi-script OCR that handles mixed English and Indian scripts on the same document, or vision-language question answering for Indian educational content.

4. Bharat Data-to-Insight Pipeline

Build a pipeline that turns raw Indian data into natural-language insights, reports, or recommendations using Gemma. Work with public Indian datasets: Census, NFHS health data, IMD weather, RBI economic indicators, Agmarknet crop prices, government schemes, election data, or municipal open data. Let Gemma analyze patterns, generate summaries, answer questions in plain language, or produce localized reports.

Push further: build a farmer's advisory that combines weather + crop price data, a citizen scheme-discovery tool that reads eligibility rules, or a health equity dashboard that surfaces district-level disparities.

Focus Areas

1. Education: tutoring, accessibility, language learning, study assistance

2. Health and Wellbeing: mental health, fitness, medical info access, elder care

3. Productivity and Daily Life: task management, finance, local services, civic tech

4. Open Innovation: anything that solves a real problem and does not fit the above

Submission Requirements

Every submission must include:

• Working demo: hosted app, runnable Kaggle Notebook, or recorded interactive session

• Public code repository: GitHub or Kaggle Notebook, well documented

• Kaggle Writeup: problem, approach, architecture, how you used Gemma, and limitations

• Demo video (3 minutes or less): public on YouTube, showing the problem and the solution in action

Gemma must be the primary inference engine for the core feature. Wrapping a proprietary API and labelling it "Gemma-powered" does not count.

Timeline

All times IST.

• From today (registration live): Register on Kaggle, form your team, and start building right away. You do not need to wait for a kickoff.

• 9 days of build time to design, code, test, and prepare your submission.

• 13 July, 10:00 AM: Submission deadline (hard cutoff, enforced by Kaggle).

• 13 July to 14 July, morning: Judging window.

• 14 July, 6:00 PM: Live online finals. Top 3 teams pitch (5 minutes demo + Q&A). Open to all, invitation link will be shared publicly.

• 14 July, 8:00 PM: Winners announced.

Prizes

Non-monetary prizes only. All submitters get a digital participation certificate and a listing on the Kaggle hackathon leaderboard.

• 1st Place: GDG Swags, Printed Certificate, Books, Stickers

• 2nd Place: GDG Swags, Printed Certificate, Stickers

• 3rd Place: Printed Certificate, Stickers

Co-Hosted By

• GDG on Campus NIT — Samrat Talukdar

• GDG on Campus GNIT — Nazreen Imam

• GDG on Campus UEM — Mahulee Lahiri

• GDG on Campus GCECT — Shubhodeep Mondal

Judges

• Saugata Sarkar — Developer at Hyland. Former Research Intern at IIT Kharagpur and ISI Kolkata (CVPR Unit). Former Project Management Intern (R&D) at Hyland.

• Nihal Gazi — AI Researcher, Developer, and Founder of KindSynapse (India). Works on CPU-efficient neural architectures, 1-bit weight quantization models, and novel ML solutions. His work focuses on stripping away computational bloat to build AI that operates elegantly under strict physical constraints.

• Sitam Meur - AI Engineer @Daily Dose of Data Science | Former AI/ML Studio Publisher @Lightning AI | GSoC Contributor @RUXAILAB | Transforming Ideas into AI Solutions 🚀Computer Science

Kaggle Competition Page

Register and start building here: https://www.kaggle.com/competitions/gemma-for-bharat-gd-go-c-jis-university-2026