projects

Project

Rift IQ

Professional LoL analytics with ML win prediction and AI coaching.

Rift IQ screenshot 1

Overview

Rift IQ moves beyond basic stat-tracking by implementing a real-time data pipeline that benchmarks player performance against the specific context of their matches. By comparing stats directly to the other nine players in a lobby rather than global averages, it provides a high-fidelity view of a player's actual impact on the game outcome.

The platform features a custom-trained XGBoost model for win probability estimation, including a specialized "Streamer Mode" that uses intelligent imputation to handle private profiles. This is complemented by a deep duo analysis system that scouts synergy between premade partners, tracking shared winrates and identifying optimal champion pairings.

To provide actionable growth paths, Rift IQ integrates with Groq to deliver personalized coaching tips. The system analyzes performance deltas across vision, combat, and economy, generating natural language advice that addresses specific weaknesses. All match history and performance trends are persisted in a local SQLite database, powering interactive LP graphs that visualize the player's competitive journey.

Stack

React, FastAPI, Python, XGBoost, SQLite, Groq, LLaMA 3.3, Riot API, Tailwind CSS, Recharts

Status

Complete