Summary
Amazon's Silicon and Systems Group is seeking a Systems Development Engineer for their Edge AI Platform Infrastructure team. This role focuses on building and maintaining the test and release infrastructure for custom AI accelerator silicon IP, kernel drivers, and ML inference software. You will design hardware-in-the-loop test frameworks, develop CI/CD pipelines, and leverage AI for automated log analysis and test optimization. This position is for engineers with a deep understanding of hardware IP and a passion for creating scalable, reliable automation.
Description
Join the Edge AI team within the Silicon and Systems Group (SSG) at Amazon. As a Systems Development Engineer, you will own the
test and release infrastructure that validates custom AI accelerator silicon IP, kernel drivers, and ML inference software across Amazon's edge device portfolio.
This role demands someone who understands hardware IP at the register level AND can build scalable automation around it. You will design hardware-in-the-loop test frameworks that exercise AI accelerator compute cores, DMA engines, and security subsystems on physical silicon — then build the CI/CD pipelines and AI-assisted workflows that make those tests run continuously, reliably, and at scale.
You won't just run existing tests. You will create new test processes from scratch — defining what to measure, building the execution infrastructure, instrumenting coverage, and using AI to
multiply your effectiveness. When an inference failure surfaces in production, you need the IP-level understanding to trace it from a CloudWatch metric through the kernel driver down to specific hardware accelerator behavior.
Key job responsibilities
Hardware-in-the-Loop Test Infrastructure: Design, build, and maintain automated test systems that execute on physical AI accelerator silicon across device pools. Own scheduling, device provisioning, firmware flashing, result collection, and failure triage at scale.
IP-Level Technical Depth: Develop deep understanding of AI accelerator hardware IP (compute cores, DMA controllers, MMU/IOMMU, power domains) to write meaningful validation
that exercises real silicon behavior, not just software interfaces.
Test Process Creation: Define and implement new test methodologies and frameworks from the ground up. Build systems that are robust to device failures, scale across multiple platform generations, and produce actionable signal.
AI-Powered Workflows: Apply AI fundamentals to force- multiply engineering effectiveness — automated log analysis, intelligent test selection, failure pattern recognition, AI-assisted code generation, and agentic CI workflows.
Coverage & Quality Tooling: Implement and maintain code coverage, test coverage mapping, Static Code Analysis (SCA), and Dynamic Code Analysis (DCA) integrated into the development workflow and CI gates.
CI/CD Pipeline Ownership: Build and maintain release pipelines using AWS services (CDK, Lambda, CodePipeline) and embedded build systems (Yocto/BitBake) that take validated
software from commit to production.
Production Monitoring: Create dashboards and alerting for on-device inference metrics, crash rates, and performance regressions across the device fleet. Investigate issues
spanning hardware accelerator behavior through kernel driver to userspace.
Cross-Team Collaboration: Work with silicon validation, platform software, ML science, and QA teams to identify test gaps, define coverage targets, and close them systematically.
