📌 About the Event

The Build with Gemma Hackathon – Nagpur is a highly focused, single-day technical building sprint organized by GDG Cloud Nagpur.

This hackathon centers completely around Google DeepMind’s open-weights model family, Gemma. The event transitions developers away from proprietary, black-box APIs to focus on local optimization, open-source model customization, and fine-tuning. Participants will collaborate to engineer next-generation applications spanning Generative AI, autonomous agents, and multimodal systems targeted at real-world societal issues.

📅 Critical Timelines & Infrastructure Gates

• Date: Saturday, July 18, 2026
• Time: 11:00 AM – 4:30 PM IST (5.5-Hour Accelerated Rapid Prototyping Sprint)
• 📍 Venue Hub: Location Pin (TBA), Nagpur, Maharashtra – 440001, India.
• 🗳️ Submission Pipeline: All software codebases, documentation, and operational artifacts must be officially hosted and submitted through Kaggle for runtime verification and testing.

⏱️ Technical Hackathon Flow

• 11:00 AM – 11:30 AM | Check-In, Environment Verification & Opening Brief Hardware setup, sandbox configuration checks, and an architectural overview of the Gemma model family capabilities.
• 11:30 AM – 03:00 PM | The Code Sprint: Building with Gemma The core development block. Teams build, fine-tune, or prompt-engineer their localized solutions with active, continuous peer networking.
• 03:00 PM – 04:15 PM | Kaggle Deployment, Team Demos & Live Evaluations Pushing finalized repositories to Kaggle, demonstrating live working environments to the judges, and collecting critique.
• 04:15 PM – 04:30 PM | Evaluation Wrap-up, Next Steps & Closing Ceremony Final feedback compilation, overview of community tracks, and group photos.

🏗️ Targeted Build Tracks & Key Pillars

The judging panel will score builds explicitly on how well they leverage Gemma’s lightweight architecture across these domains:

• 🤖 Agentic AI & Planning: Building self-correcting agent structures, tool-calling pipelines, and autonomous execution loops.
• 🌐 Local Language Solutions: Implementing Low-Rank Adaptation (LoRA) or fine-tuning parameters to adapt Gemma for regional Indian languages and dialects.
• 👁️ Multimodal Systems: Combining diverse input types (text, code, structured data charts) to generate high-accuracy outputs.
• 💡 Civic & Impact Engineering: Deploying lightweight models locally to solve accessibility, public education, healthcare data triage, or productivity challenges.

🏆 Evaluation Metrics & Rules

• Effective Gemma Implementation: Submissions must demonstrate core utilization of the Gemma open model family rather than standard API wrapper calls.
• Functionality & Impact: The application must compile and perform live inference successfully during the judging demo.
• Originality: Projects must be built freshly during the sprint windows; pre-existing templates or plagiarism trigger instant platform disqualification.