Senior Manager, AI Engineering, Amazon Leo, D2D Software

Description

Amazon Leo is building a low Earth orbit satellite constellation to bring fast, affordable connectivity to customers around the world. Direct to Device (D2D) extends Leo to standard mobile phones and devices: a fully regenerative fifth-generation (5G) mobile base station on each satellite, delivering service to standard handsets through mobile network operators (MNOs). D2D Software builds the embedded software that flies, the cloud and ground software that runs the service, and the platform that connects them. Our target is under a day from a defect being found to the fix deployed across the entire constellation, achieved by automating everything that a person would otherwise do by hand.
This role leads AI engineering in DevOps & Release, reporting to the Director, DevOps & Release. It owns triage, test generation and selection, and campaign monitoring across the platform. Before anything is handed on, the bar is: each AI stage meets its acceptance criterion, with a manual fallback. It hands off to its customers as follows: triage, test generation, and selection to every team.
What the team needs to achieve
• Each AI stage meets its acceptance criterion, with a manual fallback.
• Hand-offs that meet their service level: triage, test generation, and selection to every team.
• Measured, rising capability and efficiency of the team, with manual work falling release over release.
How the team is measured
• Each AI stage meeting its acceptance criterion.
• Gap between the serial loop budget and the pipelined loop time.
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
• Own triage, test generation and selection, and campaign monitoring across the platform.
• Meet the bar before every hand-off: each AI stage meets its acceptance criterion, with a manual fallback.
• Work directly with engineers in the boundary teams as partners, through agreed contracts and service levels.
• Automate everything a person would otherwise do twice.
• Hire, coach, and grow the team, with clear expectations and career development for every engineer.
• Define the team's metrics and service levels, and improve them release over release.
• Make the team's decisions data-driven and its quality measurable, and escalate early when risks cross team boundaries.
• Set the team's roadmap and the structure and mechanisms that deliver it, and negotiate the boundaries of the area with peer teams.
• Build the mechanisms and best practices that make the team's work repeatable, measurable, and improving.
A day in the life
About the team
D2D Software is an evolving organization of many teams: Software Architecture & Specifications; Software Delivery & Business Ops; Software Integration, Verification & Reliability; DevOps & Release; Satellite Software Development; Ground Software Development; and Customer Software Engineering & Support. Each Director owns business outcomes: delivering the commitments, and the capability and efficiency of their teams. Teams work as internal customers and suppliers with agreed service levels, interfaces are agreed centrally while designs stay local, and engineers are expected to work directly with engineers across every boundary. Milestones are committed deadlines for delivery with quality, not gates to wait on, and every team runs on one toolchain and one system of record. Every measure is derived from the systems of record, never self-reported, and we improve release over release with a target of perfection.