At AVIAN, our mission is clear: we stop fires before they start. Our advanced infrared camera systems and AI-driven monitoring solutions help manufacturers protect people, assets, and the environment from devastating fire risks. Partnering with global insurers and industry leaders, we deliver technology that safeguards critical operations and builds a culture of safety across the wood, manufacturing, and energy sectors.

Why this role matters:

This isn’t research for slides—it’s mission-critical engineering. You’ll design, build, and deploy the algorithms that keep entire facilities safe. When our system fails, real operations are at risk. If that pushes you rather than deters you, this is your place.

What you’ll do:

• Engineer and optimize CV/ML pipelines in Python and C++ for real-time inference on embedded and Linux-based edge devices.
• Deploy at scale on embedded hardware and in the cloud.
• Fuse multimodal data (thermal, RGB, signal-based) and integrate LLM-powered models for detection, scoring, and anomaly analysis.
• Own the lifecycle: from early-stage research to production rollout, monitoring, and post-deployment improvements.
• Raise the bar: conduct reviews, mentor teammates, and set technical direction across the CV/ML stack.

How success is measured:

• Real-time performance meets or exceeds production KPIs (latency, precision, recall).
• Stable AWS deployments with zero-rollback production updates.
• Documented model improvement metrics across successive releases.
• Measurable decrease in alarms caused by nuisance phenomena like exhausts, chainsaws, welding, etc.
• Field devices operate autonomously for >30 days with zero human intervention.

• Replicate one end-to-end CV pipeline from repo to edge deployment.
• Profile thermal and RGB models on real hardware; benchmark latency and memory.
• Deliver one performance or inference optimization patch.

• Take ownership of a core subsystem (motion analysis, temporal filtering, or anomaly detection).
• Deploy your subsystem via AWS ECS + Lambda orchestration.
• Provide reproducible benchmarks showing measurable improvement.

• Lead a live production rollout and post-deployment validation.
• Implement automated monitoring with error metrics and alerting.
• Deliver a concise design doc or internal paper establishing new performance standards.