EagleEye
Multimodal roadway image analysis with structured incident alerts and response recommendations.
Overview
A roadway image contains useful incident context, but an application needs consistent, structured information to act on it. The engineering problem is to bridge visual interpretation and a predictable software interface.
EagleEye uses the Claude API to analyze roadway camera images and return JSON with incident details and recommended police, fire, EMS, or towing resources.
How it works
- Roadway camera images provide the multimodal input.
- The Claude API produces structured incident information and response recommendations.
- The service integrates AWS Lambda, API Gateway, Amazon S3, and CloudWatch.
Technical implementation
- Input validation checks requests before analysis.
- Retries and timeouts handle failures and delays in AI API calls.
- Structured JSON gives downstream consumers a consistent response format.
- Cloud logging supports debugging across service integrations.
Model interpretation can be uncertain. Recommended resources are model outputs, not verified dispatch decisions. Production deployment, real emergency dispatch, and detection accuracy are not established by the supplied material.
Results
1st Place in the Open Track at the Claude Builders Club @ OSU Hackathon. Built an end-to-end multimodal service returning structured incident alerts.
The resume documents input validation, retries, timeouts, and logging as reliability techniques. A test suite, incident dataset, and measured detection results were not supplied.