Software Development Engineer, Ads DTx Bifröst

Are you looking for a Software Development Engineer position with Amazon-scale impact, on a team that has already re-tooled how it builds? Advertising at Amazon is growing fast by giving vendors and sellers efficient tools to drive sales. Our advertisers need tools that help them get the most from their advertising dollar in the least time. We provide a world class experience to hundreds of thousands of advertisers globally, helping them grow their businesses and run more successful campaigns. Our team owns the console framework and portal services that every Amazon Advertising surface is built on: authentication, advertiser and account context, navigation, and the routing and platform capabilities that the rest of the advertising org depends on. What sets this role apart is how we work. We are a native AIDLC (AI-Driven Development Life Cycle) team. We use agentic development across the full cycle, from design and requirements through implementation, review, testing, and operations, and we improve the practice as we go. We are not experimenting with AI on the side. It is how we ship. We are looking for an engineer who wants to build customer-facing platform services and, at the same time, help define what high-velocity agentic engineering looks like at Amazon. You will own the design and implementation of major deliverables end to end. You will care about quality, consistency, maintainability, performance, and security, and you will use agentic tooling to hold a higher bar on all of them rather than a lower one. This position requires collaboration with software engineers, product managers, technical program managers, and partner teams, along with comfort in ambiguity and a drive to create, iterate, and improve. Key job responsibilities - Own the design and delivery of platform capabilities on the ads console framework and portal services, end to end through production, for hundreds of thousands of advertisers. - Work agentically across the full development life cycle: use AI to draft designs, scaffold implementations, generate and review code, expand test coverage, and investigate operational issues, while staying accountable for correctness and judgment. - Drive ambiguous problems to simple, data-backed decisions, including one-way-door architecture choices, and write designs at planning altitude that partner teams can align to. - Set and hold the operational and quality bar: test coverage gates, security event logging, threat models, alarms, and safe, reversible rollouts. - Improve our AIDLC practice itself. Codify what works into reusable prompts, agents, skills, and harness workflows so the whole team gets faster and more reliable over time. - Collaborate across teams and partners, and influence shared platform decisions. A day in the life You own a capability, not a ticket queue. A typical day is a mix of deep design work and driving agentic workflows that do the mechanical heavy lifting under your direction. You might start by refining a design with an agent, pressure-testing assumptions and trade-offs before you commit to a one-way door. You dispatch agents to scaffold the implementation, sweep the codebase for every call site a change touches, and draft the tests, then you review their output with the same rigor you would apply to a peer's code review. When an alarm fires or a build breaks, you lean on the harness to triage and localize before you decide the fix. Through the day you spot friction in how the agents work and turn it into a reusable skill or workflow, so tomorrow is faster than today. You still make the calls that matter: the architecture, the correctness bar, the customer trade-off, and the decision to ship. The team operates in single-engineer pods with an agentic harness, so you have real ownership and real leverage. You are expected to behave as a force multiplier, not a single-threaded contributor.

We are Bifrost, the team behind Portal Server and the console framework that Amazon