Applied Scientist, Amazon Optics

Description Amazon Optics owns the design, delivery, and operation of the unified physical security platform that protects Amazon's global infrastructure. We consolidate alarm management, device monitoring, and video surveillance data to deliver real‑time situational awareness and actionable insights for security professionals. We work on complex challenges spanning automated threat detection, signal processing, and AI‑driven security evaluation, and we are looking for talented people who want to help protect Amazon's people, assets, and operations at scale.

Come work for the Optics Incident Response Tech team, specializing in the detection, triage, and resolution of physical security incidents across Amazon's global footprint. We are looking for an Applied Scientist who is passionate about applying artificial intelligence and machine learning to physical security operations. You will build services and tools that leverage AI/ML to automate false alarm resolution, enable intelligent signal processing, and deliver predictive insights from security event data.

You will be part of a team building Large Language Model (LLM)-based services focused on enabling security teams to interact with alarm, access control, and video surveillance data in real time. The team works in close collaboration with other Optics services including Signals, Access Control Management, Video Surveillance and Analytics to power automated mitigations that protect Amazon's global infrastructure.

• Design and develop ML models for automated alarm triage, false alarm suppression, and anomaly detection across physical security signal streams (access control events, intrusion alarms, video analytics triggers)
• Build and optimize LLM-based systems that enable security operators to query, summarize, and interact with incident data using natural language
• Develop predictive models that identify emerging security patterns, correlate multi-source signals, and enable proactive incident response before threats materialize
• Research and implement computer vision applications for video-based threat detection, object classification, and automated situational awareness during active incidents
• Architect end-to-end ML pipelines from data ingestion through model training, evaluation, deployment, and monitoring in production
• Collaborate with software engineers to integrate ML solutions into the Optics platform's real-time event processing infrastructure
• Define evaluation frameworks including precision/recall metrics for alarm classification, latency requirements for real-time inference, and A/B testing protocols for model improvements
• Own model performance in production including drift detection, retraining cadences, and incident-driven model updates
• Identify and scope ambiguous problem areas where ML can transform incident response workflows, even when the business problem is not yet fully defined
• Drive scientific breakthroughs in areas such as multi-modal fusion (combining video and sensor data), few-shot learning for rare security events, and reinforcement learning for dynamic response prioritization
• Advise team members on ML best practices, model selection, feature engineering, and experimental design
• Influence product roadmaps by translating scientific capabilities into customer-facing features that reduce mean time to resolution and false alarm rates
• Engage with cross-functional stakeholders including security operations, hardware engineering, and partner teams (Signals, Access Control Management, Analytics) to align scientific investments with operational needs
• Reduce false alarm rates through intelligent signal classification and contextual filtering
• Decrease mean time to detection and resolution for genuine security incidents
• Increase automation coverage for routine incident types, freeing human operators for complex threat assessment
• Publish internal research and contribute to the broader ML/security community through papers and patents

• 3+ years of building models for business application experience
• PhD, or Master's degree and 4+ years of CS, CE, ML or related field experience
• Experience in patents or…