Skip to content

oluchi-i/clairvoyant-workers

Repository files navigation

CLAIRVOYANT

AI-powered race walking rule violation analyzer built on Cloudflare Workers AI

🔴 Live Demo: https://my-ai-app.oluchi-ai.workers.dev


What It Does

CLAIRVOYANT analyzes race walking athlete movement descriptions and automatically detects rule violations based on World Athletics Laws, the same two rules enforced in Olympic competition:

  • Loss of Contact: both feet leaving the ground simultaneously
  • Bent Knee: advancing leg not straightened at ground contact

The app returns a structured verdict: RED CARD, WARNING, or NO VIOLATION, with the specific rule violated, confidence level, reasoning, and recommended judge action.


Why I Built This

This app extends a Senior Research project, also named CLAIRVOYANT, where we built a Python/OpenCV computer vision pipeline that analyzed race walk athlete footage frame-by-frame to detect violations automatically, achieving a high accuracy on the primary detection class.

This Cloudflare Workers version brings that same domain expertise to a serverless AI platform, replacing the CV pipeline with a natural language interface powered by Cloudflare Workers AI (Llama 3.3 70B).


Tech Stack

  • Cloudflare Workers: serverless edge deployment
  • Cloudflare Workers AI: Llama 3.3 70B instruct model
  • TypeScript: type-safe Worker implementation
  • Wrangler: CLI deployment and local development

How to Run Locally

git clone https://github.com/oluchi-i/clairvoyant-workers
cd clairvoyant-workers
npm install
wrangler dev --remote

Visit http://localhost:8787


How to Deploy

wrangler deploy

Example Input & Output

Input:

The athlete's right foot leaves the ground as their left foot is still in the air, resulting in a brief moment where neither foot is in contact with the ground. The knee remains bent at ground contact.

Output:

VERDICT: RED CARD
RULE VIOLATED: Loss of Contact
SECONDARY VIOLATION: Bent Knee
CONFIDENCE: High
REASONING: The athlete clearly violates the Loss of Contact rule by having both feet off the ground simultaneously. The bent knee at ground contact constitutes a secondary violation of the Bent Knee rule.
RECOMMENDED ACTION: The judge should immediately issue a red card to the athlete.

Related Project

See the full computer vision implementation of CLAIRVOYANT: ...


Built by Oluchi Ikwuegbu: Drexel University CS 2026

About

~ Race Walk Violation Detector

Topics

Resources

License

Stars

0 stars

Watchers

0 watching

Forks

Releases

No releases published

Packages

 
 
 

Contributors