Senior Software Development Engineer, Amazon Quick

As AI agents take on more autonomous actions on behalf of users, a critical question emerges: how do you ensure every one of those actions is authorized, auditable, and safe? That is the problem you will help solve on the Amazon Quick access-control team.

As a Software Development Engineer, you will design, build, and operate the authorization policies, enforcement plane, and audit systems that govern every interaction in Amazon Quick, whether initiated by a person or by an AI agent acting on their behalf. Your work will directly enable Amazon Quick to scale its agentic capabilities while meeting the access-control requirements that enterprise customers depend on. You will collaborate across more than a dozen partner teams, influence architectural decisions in systems you depend on, and coach other engineers on your team. This is an opportunity to shape how the industry approaches access control for agentic AI, at a time when the problem is still being defined.

Key job responsibilities

• Design and implement authorization policies, enforcement mechanisms, and audit pipelines that ensure every human and agent interaction in Amazon Quick is governed by a correct, enforceable access decision.
• Lead the architecture of distributed access-control systems, balancing scalability, security, and maintainability while identifying and resolving risks before they reach production.
• Collaborate with partner teams across the Amazon Quick organization to align on integration points, define service contracts, and ensure access controls keep pace with new agentic capabilities.
• Evaluate and integrate generative AI tooling into development workflows where it accelerates delivery, and apply responsible practices when reviewing AI-generated outputs.
• Mentor and grow engineers on the team through code reviews, design guidance, and technical coaching, raising the quality bar for the team"s software practices.

A day in the life

You might start the morning reviewing an access-policy change proposed by a partner team, verifying it meets audit requirements before it rolls out. After a design review with your team, you spend focused time prototyping a new enforcement mechanism that handles delegated agent permissions. In the afternoon you pair with a teammate to debug an edge case in the audit pipeline, then wrap up by writing a short design document that captures the trade-offs you explored. Your work spans both solo deep-thinking time and collaborative sessions with engineers across the Amazon Quick organization.

About the team

Our team owns the authorization and access-control layer for Amazon Quick, making sure the product is enterprise-ready for the world"s largest organizations. We partner with teams across the full Amazon Quick organization, over 1,000 engineers, giving you broad visibility into how the product evolves. As agentic AI grows more capable, the access-control challenges we tackle are new and still being defined across the industry, so you will have real influence over the direction we take. If you are energized by security-critical problems at scale and want to work where AI meets enterprise trust, this is the team for you.