Epic moment reels

Finds your best plays, comebacks, and clutch fights from LoL match history and turns them into a shareable 60-second highlight reel, narrated and captioned by AI and produced entirely on AWS.

Comment

Inspiration

Players love reliving their best League moments, but existing platforms don’t tell the emotional, cinematic story behind the stats. I wanted to create a quick, fun “highlight reel” experience powered by AI.

What it does

Epic Moment Reels turns a player’s match history into a personalized recap with:

  • Detected highlights (pentakills, carries, comebacks)
  • AI-generated captions
  • A TikTok-style scrolling reel UI
  • A short “season story” summary

How I built it

  • Rails backend for Riot API calls and highlight detection
  • Rule-based engine for identifying epic plays
  • OpenAI for generating captions + narrative
  • React frontend for a vertical reel interface
  • Entirely local, no cloud infrastructure

Challenges I ran into

  • Normalizing inconsistent Riot match data
  • Designing simple but meaningful highlight rules
  • Keeping the LLM prompts structured and predictable
  • Creating a UI that feels “reel-like” without building real video

Accomplishments that I'm proud of

  • End-to-end experience built without any heavy infra
  • Super fast local dev loop
  • Clean highlight detection engine
  • Reel UI that feels engaging and social-ready

What I learned

  • Riot’s match data is deep but tricky to stitch together
  • LLMs work best with tightly structured JSON prompts
  • You can get 80% of the “video recap” feeling with just UI and storytelling

What's next for Epic Moment Reels

  • Add real video export using local FFmpeg
  • More advanced highlight rules (objective steals, lane dominance, gold swings)
  • Add comparison reels between friends
  • Upgrade to AWS Bedrock for a cloud-ready version
Share this project:

Updates