About the Team

Amazon's Devices & Services Trust CX Innovations team is building the future of responsible AI for consumer devices. We deliver privacy-first, accessible, and trustworthy AI experiences that millions of customers use daily across Amazon's device ecosystem—including Alexa, Echo, and other ambient computing products. Our mission is to ensure that as we push the boundaries of generative AI innovation, we maintain Amazon's high bar for customer trust, privacy, inclusion, and accessibility.

Role Overview

We are seeking an exceptional Software Development Engineer III to build the foundational systems and consumer-facing features that enable trustworthy AI experiences at scale. You will partner closely with our Product Manager-Technical to design and implement privacy-preserving AI architectures, responsible AI frameworks, and accessibility features that set the standard for consumer AI products.

This is a high-impact role where you'll tackle complex technical challenges at the intersection of AI innovation and customer trust—balancing performance with privacy, building explainable AI systems, and creating guardrails that protect customers while enabling delightful experiences.

What You'll Build

Privacy-Preserving AI Systems:

* Design and implement federated learning and differential privacy techniques for on-device and hybrid AI architectures
* Build privacy-preserving data pipelines that minimize data collection while maintaining model quality
* Create consent management frameworks and data minimization systems that give customers control
* Architect on-device vs. cloud processing trade-offs that optimize for both privacy and performance

Responsible AI Infrastructure:

* Develop AI evaluation frameworks to measure model quality, safety, and bias across diverse customer populations
* Implement guardrails and safety boundaries for LLMs and foundation models in consumer applications
* Build observability and monitoring systems for AI performance, hallucination detection, and trust metrics
* Create human-in-the-loop review systems and escalation pathways for edge cases

Accessible & Inclusive AI Experiences:

* Implement WCAG 2.1 AA and Section 508 compliance for AI-powered interfaces across voice, visual, and multimodal experiences
* Build accessible AI features that work seamlessly for customers with diverse abilities
* Design inclusive AI systems that perform equitably across different demographics and use cases

Consumer-Facing Trust Features:

* Develop explainable AI interfaces that help customers understand how AI makes decisions
* Build transparency controls that show customers what data is used and how
* Create privacy dashboards and settings that give customers meaningful control over their AI experiences
* Implement cross-device identity management with privacy-first design principles

Key Technical Challenges

* Latency vs. Privacy Trade-offs: Optimize time-to-first-token and response times while maintaining strong privacy guarantees through on-device processing and selective cloud offloading
* AI Safety at Scale: Reduce hallucinations to