Astra Coach

AI assistant coach transforming raw esports match data into actionable coaching insights. It analyzes player mechanics, team strategy, detects tactical mistakes, and generates post-match summaries.

Comment

Inspiration

Professional esports teams generate massive amounts of data, but coaches still spend hours translating statistics into meaningful decisions. We built Astra Coach to act like a real assistant coach — not just showing numbers, but explaining why things failed, what worked, and what to fix next.

Inspired by Moneyball-style analysis and real coaching workflows, Astra Coach bridges the gap between analytics and strategy.

What It Does

  • Astra Coach provides:
  • Micro analysis: Player mechanics, consistency, ability efficiency, and positioning
  • Macro analysis: Rotations, objective timing, coordination, and recurring team patterns
  • Mistake detection: Automatically identifies repeated tactical and positional errors
  • AI coaching summaries: Clear, actionable post-match feedback written in professional coaching language
  • Interactive dashboard: Visual insights for coaches and analysts

How we built it?

  • JetBrains PyCharm for backend development
  • Junie (AI Coding Agent) to:
  • Generate analytics modules
  • Design mistake detection logic
  • Create AI coaching prompt templates
  • Refactor, document, and validate code
  • Python + FastAPI for analytics and APIs
  • Pandas / Numpy for data processing
  • Plotly for visual analytics
  • Streamlit for the interactive coaching dashboard

All data pipelines, logic, and AI prompts are fully transparent and explainable.

Challenges We Faced

  • Designing mistake detection that feels coaching-relevant, not generic
  • Translating raw metrics into language coaches actually use
  • Balancing realism with hackathon time constraints
  • Ensuring AI outputs are actionable, not vague

Accomplishments that we're proud of

  • Built a fully working coaching assistant, not just a dashboard
  • Generated human-like coaching summaries with clear tactical intent
  • Integrated Junie meaningfully into real development workflows
  • Created a system coaches could realistically use post-match

What we learned

  • Explainability matters more than model complexity
  • Coaches value clarity and context over raw numbers
  • AI works best as a partner, not a replacement, in decision-making

What's next for Astra Coach

  • Live match ingestion
  • Game-specific logic (LoL / VALORANT)
  • Practice planning recommendations
  • Exportable coaching reports (PDF)
  • Multi-match trend analysis

Built With

Share this project:

Updates