Sr. Software Dev Engineer, AWS Insights and Optimizations

Every AWS customer wants to know the same thing: am I paying for more cloud than I actually need? AWS Insights and Optimizations is the team that answers that question, for millions of accounts, across compute, storage, databases, and commitment purchases. We analyze resource insights, compute recommendations against the latest AWS offerings and pricing models, and build the tooling that automates the optimization workflow end to end.

That question has gotten harder and more expensive to get wrong. AI and machine learning workloads are now the fastest growing line on many customers" bills. Accelerated compute is scarce and costly, training and inference usage is bursty and hard to forecast, and the usual rules of thumb for whether a resource is right-sized do not transfer cleanly to a GPU fleet or an inference endpoint. Customers want the same clarity on their AI spend that they have come to expect on the rest of their infrastructure, and today most of them do not have it. Closing that gap is a significant part of where this team is going.

You would own the foundation the whole experience is delivered on. Customers come at optimization from a lot of directions: the console, our public APIs, a partner product, an export into their own reporting stack, a conversation with their support team, or an agent asking on their behalf. Every one of those paths runs through the platform in this role, so the work is a customer experience problem before it is an infrastructure problem. A finance lead needs to see an entire organization at once rather than one account at a time. An engineer deciding whether to accept a change needs the answer quickly, needs it to reflect this week"s usage and not last quarter"s, and needs to understand how we reached it before they will act on it. A partner or a customer"s own automation needs to build on us and still work a year from now.

Who we serve is also changing. The experience was designed for a person clicking through a console, and it is increasingly driven by agents that run continuously, ask in bulk, and need to know not only what we recommend but why. We treat that as a build mandate rather than a buzzword. You would help redesign the optimization experience around generative AI: opening the platform to agents through MCP and agent-friendly contracts, carrying enough provenance with every recommendation that an agent can justify an action to the human accountable for it, supporting multi-step and batch work instead of one question at a time, modernizing authentication for machine-to-machine scale, and making the system behave under continuous, bursty, non-human load. You would use the same tools on your own work. Agents are already part of how this team builds and operates, and we would want you pushing that further.

If you like owning a broadly used API surface, care about the difference between a recommendation that is technically correct and one that earns a customer"s trust, and want your work to show up as real dollars off real bills, come talk to us.

Key job responsibilities

• Own the public API surface customers, partners, and other AWS services depend on to enroll, retrieve recommendations, export data, and express how they want their environment optimized, including the cross-account and organization-wide access model behind it.
• Design for agents as first-class consumers: programmatic and MCP access, bulk and multi-step operations, machine-to-machine authentication, and capacity and throttling behavior that holds up under continuous automated load.
• Extend the platform to AI and accelerated compute spend, including the question of what optimal even means for workloads whose usage patterns look nothing like a traditional server fleet.
• Make recommendations explainable and self-diagnosable, so that customers and support teams can answer why a recommendation says what it says, or why it is missing, without an engineer in the loop.
• Raise the operational bar: how we monitor, test, deploy, and expand across Regions and partitions, and how much of the recurring operational load we can automate away instead of absorbing.
• Grow the engineers around you. Review designs, mentor, and write the documents that build consensus with stakeholders across organizations.
• Work directly with product managers and with the teams building on your APIs to decide what to build, proactively identify what is blocking the team and what the platform will need as usage grows, before either becomes a constraint.

A day in the life

You might spend the morning in a design review arguing whether a new capability belongs in the public API, then pair with an engineer on failure handling in a cross-account delivery path, then dig into a signal that says customers are enrolling but encountering issues in the optimization journey, then point an agent at your own API to find out how badly it reads to a non-human caller. Some weeks are heads-down implementation. Some weeks are writing the document that decides the next six months.

About the team

AWS Insights and Optimizations is a group of engineers who like problems that sit between large-scale data processing and customer-facing APIs, and we are unusually close to the outcome of our work, because we can measure it in savings customers actually realized. We are among the more fastest-moving adopters of generative AI in AWS cloud financial management, in the product and in how we build and operate it. We care about operational excellence, we write things down, and we would rather have the hard architectural conversation early than live with the shortcut for three years. The charter is expanding and there is real room to shape it.