Software Development Engineer, AI Foundations, Amazon Leo AI Foundations

Description

Amazon Leo is Amazon's Low Earth Orbit satellite network, built to deliver fast, reliable internet to customers beyond the reach of existing networks. The AI Foundations team builds AI agents that turn Leo's best troubleshooting expertise into diagnoses in minutes instead of hours, helping keep the network available for every customer.
As a Software Development Engineer on AI Foundations, you will design and build production agentic systems on AWS that reason over real satellite, ground, and network data. You will work directly with principal, Sr. Principal engineers and network operators, and your work will shape how Leo is operated at scale.
Export Control Requirement: Due to applicable export control laws and regulations, candidates must be a U.S. citizen or national, U.S. permanent resident (i.e., current Green Card holder), or lawfully admitted into the U.S. as a refugee or granted asylum.
Key job responsibilities
- Design, build, and operate AI agents and services that diagnose network and connectivity issues using telemetry, logs, and engineering knowledge.
- Turn proven engineering tools and data sources into reusable agent tools so teams across Leo can build on them.
- Build evaluation harnesses that measure agent accuracy, coverage, and latency, and use them to gate launches.
- Own your services in production: on-call, operational metrics, root-cause analysis, and security reviews.
- Partner with principal engineers, operators, and program managers to turn ambiguous problems into shipped features.
- Write clear design docs, give thoughtful code reviews, and help teammates adopt agentic development practices.
A day in the life
You might start the morning reviewing overnight evaluation results and spotting a class of network failures your agent misdiagnosed. You trace the gap to a missing data source, then sit down with the principal engineer who owns that subsystem to agree on how to expose their tool to the agent. After lunch you build the integration and add new test cases, then review a teammate's pull request. Late afternoon, you join an operations review where an operator walks through a real incident your agent helped resolve, and you leave with a clear next priority.
About the team
We are an AI-native team of software engineers and applied scientists in Redmond, and we move fast. We use AI in how we build, not only in what we build, so a good idea can be a working experiment the same week and a production capability the next. We think beyond how things have been done, test our ideas against real problems, and keep only what proves itself. We ship lightning fast, and we hold a high bar for quality and production readiness while doing it.