Chidi Enwereji

SiteSentinel: an AI safety and risk inspector built for Meta Ray-Ban glasses SiteSentinel turns what a worker sees into an inspection. Point the glasses, or a phone mounted on a hard hat for now, at a worksite, and it finds hazards with **SAM 3.1**. It then scores each one for likelihood and impact (out of 25) with **Muse Spark**, covering people, equipment, fire, structural and process risks. It suggests fixes in the usual order of safety controls, from removing the hazard down to protective gear. It's an agent, not just a detector. It logs each issue, assigns it to the right person on site, speaks the alert out loud and takes voice reports. It then checks an "after" photo to close the issue or reopen it, and keeps a full audit trail. Rule packs for construction (OSHA standards) and oil and gas make it work across industries. **Built with:** Meta SAM 3.1, Muse Spark (with Muse Spark Open on Baseten as a backup), Meta Voice Transcribe, and the Wearables Device Access Toolkit starter app. It runs on TypeScript and SST on AWS (Lambda, DynamoDB, S3, CloudFront).