Project Story

About the Project

RoadToPro is a real-time esports analytics platform that ingests, processes, and visualizes League of Legends match data. It combines live match feeds from multiple sources, runs machine learning predictions, and provides a React dashboard for actionable insights. The platform also integrates an AI-powered chatbot to help users understand match dynamics in real time.

Inspiration

As esports continue to grow globally, real-time insights into professional matches are in high demand. We were inspired by the challenge of turning raw, high-frequency match data into meaningful insights for both fans and analysts. Traditional dashboards often lag behind live events, and we wanted to create a platform that provides instantaneous data, predictive analytics, and AI-assisted guidance.

What it Does

RoadToPro:

Continuously ingests match data from PandaScore and LoL Esports APIs.
Streams events via Kafka/Redpanda to handle high-volume data.
Stores structured data in PostgreSQL for analysis and historical insights.
Visualizes live matches with gold difference charts, momentum graphs, and player stats.
Predicts win probabilities using machine learning models in real time.
Provides an AI chatbot for natural-language queries on ongoing matches.

How We Built It

We combined a modern full-stack architecture with event-driven design:

Backend: Python 3.12, FastAPI, AsyncIO for high-concurrency ingestion.
Data Streaming: Kafka/Redpanda for reliable event messaging.
Database: PostgreSQL for structured storage.
Machine Learning: Agents compute live win probabilities and detect highlights.
Frontend: React + Vite + TailwindCSS + Recharts for interactive dashboards.
AI Integration: Ollama LLM powers the chatbot for real-time match insights.
Infrastructure: Docker Compose orchestrates all services with Prometheus for monitoring.

We implemented asynchronous consumers, data normalization pipelines, and a modular architecture to support future games beyond League of Legends.

Challenges We Ran Into

High-frequency data: Ingesting and processing live LoL feeds without lag required fine-tuning Kafka consumers and Python async loops.
Rate limits: APIs like PandaScore have strict request limits, forcing us to implement robust throttling and caching.
Docker networking: Connecting Kafka, Postgres, and the FastAPI service across host and containers on Windows posed initial configuration headaches.
Real-time predictions: Computing win probabilities without blocking the dashboard required careful async programming.

Accomplishments That We're Proud Of

Fully real-time data pipeline from ingestion to live dashboard visualization.
AI-powered chatbot answering match questions in natural language.
Scalable architecture with Kafka + PostgreSQL + Python AsyncIO.
Interactive frontend with live charts, momentum analysis, and player stats.
Successful integration of multiple esports APIs into a unified system.

What We Learned

Advanced asynchronous Python programming with AsyncIO and asyncpg.
Event-driven architectures for high-throughput data processing.
Effective data normalization and real-time analytics.
Container orchestration and networking nuances with Docker on Windows.
Building a full-stack ML-ready platform from scratch, integrating backend, frontend, and AI models.