AI computer vision for health, safety & environment (HSE)
Leading delivery of an AI incident-reporting system that uses existing CCTV to detect safety risks in real time and trigger alerts and on-site hooters.
9+AI detection use cases
0new cameras required
Context
Industrial sites already had CCTV everywhere, but footage was only reviewed after an incident. Safety teams relied on manual patrols and paper reports to catch PPE violations, near misses and unsafe behaviour.
Challenge
Make existing cameras useful in real time: detect many kinds of risk on live video, keep false alarms low enough that people trust the system, and get the right alert to the right person, or to a physical hooter, within seconds.
What I did
- Led the programme end to end across the computer-vision, platform, integration and customer site teams, from use-case discovery to rollout.
- Planned the rollout on existing camera infrastructure, so sites needed no new cameras and could go live faster.
- Drove the use-case roadmap: near misses, PPE compliance (helmets, jackets, safety shoes), machinery zones, theft, encroachment, spillage and attendance.
- Coordinated the alerting and hooter integration with site and facilities teams, so emergencies triggered alarms on the ground.
- Ran site pilots with the customer to tune accuracy and false alarms before scaling, and tracked both as programme KPIs.
Outcome
A live AI safety layer on top of existing infrastructure. Incidents are detected and escalated as they happen, not discovered later on tape.