A League of Legends Year-in-Review dashboard that provides personalized, AI-powered performance analysis and coaching insights for players. Made for the Rift Rewind hackathon.

Live Demo: https://rift-rewind-kohl.vercel.app/

• Comprehensive Player Analysis: Fetches and analyzes your entire year of League matches using the Riot Games API
• AI-Powered Coaching: Leverages AWS Bedrock (Claude) to generate personalized, savage-yet-constructive performance reviews
• Interactive Visualizations: Beautiful data visualizations using D3.js and GSAP animations
• Real-time Chat Coach: Chat with "Ryze", an AI coach who provides personalized advice based on your stats
• Performance Benchmarking: Compare your stats against role and rank-specific benchmarks
• Champion Pool Analysis: Identifies your best champions and suggests pool optimization
• Detailed Metrics Tracking:
  • CS/min, KDA, vision score, objective participation
  • Item build tracking with inventory snapshots (early, mid, late game)
  • Monthly performance trends
  • Role distribution and champion mastery
• 30/60/90 Day Improvement Roadmap: Structured practice plans with actionable drills

• React 19 - UI framework
• React Router - Navigation
• D3.js - Data visualizations
• GSAP - Animations
• Axios - API requests
• Vercel - Deployment platform

• Python - Backend
• Flask - Web framework
• AWS Bedrock - AI/ML service (Claude 3.5 Sonnet)
• Riot Games API - Match data and player stats
• Flask-CORS - Cross-origin resource sharing
• Render - Backend deployment

Full-Stack AI Application with React frontend, Flask backend, and AWS Bedrock integration

• Decoupled Architecture: Frontend deployed on Vercel, backend on Render for independent scaling
• RESTful API: Clean separation between presentation and business logic
• Stateless Backend: Each request is independent, enabling horizontal scaling

Riot API → Match Data Processor → Statistical Analysis → AI Prompt Generation → AWS Bedrock → Structured Insights

Key Processing Steps:

1. Match Aggregation: Fetches up to 365 days of match history with pagination handling
2. Timeline Reconstruction: Reconstructs inventory from item purchase/sale/undo events
3. Statistical Computation: Calculates 40+ performance metrics including per-minute rates
4. Benchmark Comparison: Role and rank-specific performance evaluation
5. Prompt Engineering: Generates comprehensive context for AI analysis (2000+ token prompts)

• Timeline Event Processing: Parses ITEM_PURCHASED, ITEM_SOLD, ITEM_DESTROYED, and ITEM_UNDO events
• Temporal Snapshots: Captures inventory at 3 game stages (2min, midgame, endgame)
• Deduplication Logic: Handles multiple purchases of same item with count tracking
• Champion-Specific Analysis: Aggregates builds per champion for pattern recognition

• Model: Claude 3.5 Sonnet (8000 token responses)
• Prompt Engineering: Structured 8-section coaching framework with role-specific guidance
• Context Window Management: Efficient token usage with strategic data summarization
• Conversational Memory: Chat feature maintains conversation history for contextual responses

• Rate Limiting: Implements Riot API rate limit handling with exponential backoff
• Batch Processing: Fetches matches in 100-match batches with smart pagination
• Selective Timeline Fetching: Limits timeline requests to avoid API throttling (10 per batch)
• Client-Side Caching: React context stores player data to avoid redundant API calls

• Role Detection: Multi-tiered inference (API field → champion patterns → positional heuristics)
• Benchmark Normalization: Dynamic benchmark selection based on detected rank and primary role
• KDA Calculation: Handles edge cases (0 deaths = perfect KDA handling)
• Win Rate Aggregation: Per-champion, per-role, and overall calculations

• D3.js Integration: Custom SVG-based charts for champion pool and performance metrics
• Responsive Design: Media queries for mobile, tablet, and desktop viewports
• Component Architecture: Modular React components with clear separation of concerns

• React Context API: Global state for player data and insights
• Local State: Component-level state for UI interactions
• Async Handling: Loading states and error boundaries for API calls

• Modular Structure: Separation of concerns (api.py for routes, backend.py for business logic)
• Error Handling: Comprehensive try-catch blocks with meaningful error messages
• CORS Configuration: Properly configured for cross-origin requests from frontend
• Health Checks: Monitoring endpoint for deployment health verification

• RiotAPIClient: Encapsulates all Riot API interactions with regional routing
• AWSBedrockClient: Manages AI model invocation with retry logic
• MatchDataProcessor: Pure functions for statistical computation
• InsightGenerator: Prompt engineering and response formatting
• PerformanceBenchmarks: Static benchmark data with comparison algorithms

The app compares your stats against role and rank-specific benchmarks sourced from:

• LeagueMath.com
• League of Graphs
• U.GG
• League of Legends Tools

The AI coach ("Ryze") provides analysis in a unique "roast-to-coach" style:

• Brutally honest about weaknesses
• Uses gaming terminology (e.g., "diff'd", "gapped", "inting")
• Provides actionable fixes for every criticism
• Balances savage humor with genuine coaching value

Item analysis includes:

• Inventory snapshots at different game stages (early, mid, final)
• Item purchase frequency tracking
• Champion-specific build patterns
• Timeline reconstruction from match events