Professional-Grade Brawl Stars Draft Guide

• A data-driven web application that utilizes guides and analysis from professional Brawl Stars players (not just average win rates) to provide drafting advice for ranked matches in Brawl Stars.
• This tool provides recommendations based on map, mode, and pick to the player and utilizes Google's Gemini API to generate context-aware descriptions for the reasoning behind each recommendation.

Features

• Dynamic Drafting Engine: Filters 80+ Brawlers based on current meta maps and modes.
• AI-Powered Strategy Description: Uses Gemini to generate specific advice (e.g., "Why is Gus good on Shooting Star?").
• Fallback-Ensured Data Pipeline: Custom Python scripts deal with API rate limits and normalize data as needed.
• Interactive UI: Built with Streamlit for a responsive, low-latency UI/UX.

Tech Stack

• Language: Python 3.14
• AI/LLM: Google Gemini API (2.0 Flash-Lite)
• Database: SQLite3
• Data Analysis: Pandas
• Frontend: Streamlit, Custom CSS3

Handling API Rate Limits

One of the primary challenges was managing brawler descriptions for 400+ unique map/mode combinations within API rate limits.

• Developed a fault-tolerant script to handle 429 Resource Exhausted errors.
• Implemented a professional fallback strategy that fills the database with sample descriptions when the LLM is unavailable.

Installation & Setup

1. Clone the repository.

2. Install dependencies:

   pip install - requirements.txt

3. Run the script:

   streamlit run app.py