EloEcho

EloEcho is your League “Spotify Wrapped”—a one-click, AI recap that turns a year of matches into a power level, brag-worthy cards, and a fun “This or That” gold minigame.

Comment

Inspiration

We wanted players to understand how they've been playing throughout the year. We wanted them to improve their skills through our AI-powered app.

What it does

Our app analyzes players' match statistics and assigns them a rank using our custom ranking system. We then pass their statistics to our personalized AI engine to provide unique feedback that helps them improve their gameplay.

How we built it

We use AWS Lambda to house our API and data preprocessing services. We then use AWS API Gateway and Cloudfront to handle routing and caching to our Lambda API service. We use AWS Aurora DB to store players metrics and custom ranks, and use AWS S3 to store players' match history data from the Riot API. Of course, this app is powered by AI that was built using AWS Bedrock's knowledge base and inference services.

For our frontend, we use Next.js to design the face of our app. We deploy this through Vercel.

Challenges we ran into

  1. AWS Bedrock knowledge base limits how much web data it can scrape. We had to find web sources that contain a few external links and are just purely sources of truth.
  2. Handling roles and access in AWS is a problem; working through and making sure each service has the least amount of authority assigned to function properly is a challenge.
  3. Riot API limits developers to calling their endpoints. Since our apps require hundreds of match histories for different kinds of players, we had to start downloading them early on. ## Accomplishments that we're proud of
  4. Being able to utilize AWS Bedrock to generate unique feedback for players using its knowledge base.
  5. Being able to set up a serverless API service and connect that through our API Gateway and Cloudfront to enable caching and rate limiting. ## What we learned
  6. AWS S3 is a very powerful service that can store large amounts of data and is very useful for data preprocessing and machine learning.
  7. AWS Bedrock service provides easy access to models and helps in creating smart agents by utilizing its knowledge base
  8. AWS Lambda is a strong service that can help developers spin up a function in less than 10 minutes. This function scales accordingly to traffic.

What's next for EloEcho

  1. AI-generated feedback for players' movement throughout a match.
  2. AI-personalized recommendation for players' champion/ban picks during champion select.
  3. An AI-powered minion wave management feature that helps players improve their wave-clearing mechanism by providing us their gameplay video.

Built With

Share this project:

Updates