👋 Introduction
Hi! My name is Roshan, and I'm a 17-year-old developer passionate about software engineering, League of Legends, and AI. For the Rift Rewind Hackathon, I built Hextech Link which is an AI-powered League of Legends analytics platform that transforms your gameplay data into personalized insights and coaching recommendations.
Inspired by Spotify Wrapped, Rift Rewind creates beautiful yearly recaps of your gaming journey while providing an intelligent chat assistant that answers questions like "What are my top 3 champions?" or "How can I improve my gameplay?" The system delivers real-time, cost-efficient, and fully personalized insight, no more generic tips!!.
Read the technical implementation blog here: Medium
🎯 What Makes It Special
🤖 AI-Powered Coaching: Ask natural language questions and get personalized recommendations based on your actual match history
📊 Spotify Wrapped for Gaming: Beautiful yearly recaps showing your favorite champions, rank progression, and performance trends
⚡ Real-Time Insights: Advanced RAG (Retrieval Augmented Generation) architecture processes ~100 matches per player into actionable coaching insights
💰 Cost-Effective: Serverless architecture costs only ~$0.003 per user login, scaling efficiently for thousands of players
🏗️ How It Works
Phase 1: Data Pipeline
When you connect your Riot account, the system automatically fetches your match history (~6.7MB of data), processes it through AWS Lambda functions, and transforms raw gameplay data into 250-300 narrative chunks stored in a personalized knowledge base.
Phase 2: AI Intelligence
When you ask a question, the system uses semantic search to find relevant performance data, then leverages AWS Bedrock with Anthropic Claude to generate personalized coaching responses in real-time.
🧰 Tech Stack
| Frontend | Next.js 15, TypeScript, Tailwind CSS, shadcn/ui |
|---|---|
| AI & Backend | AWS Bedrock (Claude), Google Gemini, RAG Architecture |
| Infrastructure | AWS Lambda, S3, OpenSearch Serverless, API Gateway |
| Data Sources | Riot Games API (Match-V5, Summoner-V4, Champion Mastery) |
| Development | Turborepo monorepo, Vercel deployment |
💬 Example Experience
- Connect: Log in with your Riot account
- Process: System analyzes your match history automatically
- Explore: View your personalized yearly recap
- Ask: "What champions should I play to climb higher?"
- Learn: Get AI-powered coaching based on your actual gameplay patterns
🚀 Impact & Future
Current Achievement: Successfully processes League of Legends data for personalized AI coaching with sub-5-second response times
What's Next:
- 🎯 Enhanced Visualizations: Interactive performance charts and match heatmaps
- 🏆 Competitive Analysis: Compare your performance against players in your rank
- 👥 Social Features: Share recaps and compete with friends
- 🔮 Predictive Analytics: AI-powered rank prediction and improvement roadmaps
For detailed technical implementation, check out my Medium article or explore the GitHub repository.
Built With
- amazon-opensearch-serverless
- amazon-web-services
- anthropic-claude
- aws-api-gateway
- aws-bedrock
- aws-cloudformation
- aws-iam
- aws-lambda
- aws-sam
- champion-mastery-v4-api
- eslint
- etl-pipeline
- github
- google-gemini-api
- javascript
- league-v4-api
- lucide-react
- match-v5-api
- microservices
- monorepo-structure
- next.js-15
- next.js-cookies
- node.js
- pnpm
- prettier
- python
- rag-(retrieval-augmented-generation)
- react-19
- riot-games
- serverless-architecture
- shadcn/ui
- summoner-v4-api
- tailwind-css
- turborepo
- typescript
- vercel
Log in or sign up for Devpost to join the conversation.