Raikou - Hackathon Submission

Inspiration

Toyota provided several Excel files with complex race data. We saw an opportunity to make motorsport analytics accessible through natural language—letting anyone ask questions like talking to a racing expert.

What it does

Raikou is an AI chatbot that transforms thousand records of Toyota racing data into conversational insights. Ask "Who won Race 1?" or "Compare drivers 13 and 22" and get instant, accurate answers from lap telemetry, weather data, and race results. It also has an explain feature that will explain/analyze specific racing dataset and give insights regarding the context and more.

How we built it

• Frontend: Next.js 15 + TailwindCSS (Toyota Racing theme)
• Backend: Node.js/Express + MongoDB Atlas
• AI: Google Gemini 2.5 Flash (RAG pattern)
• Data: Custom parser ds
• Cost: $0 (all free tiers)

Challenges we ran into

• Inconsistent CSV formats (semicolons vs commas)
• Designing efficient schemas for diverse data types
• Teaching AI to generate accurate MongoDB queries
• Maintaining sub-5-second response times

Accomplishments that we're proud of

• 100% free infrastructure (Gemini, MongoDB, Vercel, Railway)
• Production-quality Toyota GAZOO Racing UI
• Natural language understanding for complex race analytics

What we learned

• RAG system architecture for real-time data + LLM reasoning
• Schema design for hierarchical racing data
• Prompt engineering for reliable database query generation
• Building performant full-stack apps on free tiers

What's next for Raikou

• Multi-series support (F1, NASCAR)
• Visual analytics and lap comparisons
• Predictive race insights
• Voice interface for pit crews
• Real-time race integration
• Team collaboration features

Built With

• axios
• express.js
• gemini
• mongodb
• mongoose
• nextjs
• node.js
• rag
• railway
• tailwind
• vercel