Amazon Leo is launching satellite-based connectivity in 100+ countries. Every country launch requires validation across tax, accounting, disbursement, legal, trade, and core commerce — thousands of test cases, dozens of teams, months of sequential human work. We're replacing that entire process with orchestrated AI agent squads that execute in parallel, share cross-domain context, and learn from every launch.
Leo Regulus is an AI-native team building autonomous agents that compress Amazon Leo's ecommerce launch lifecycle from months to minutes. We don't write test cases —nwe build agents that write, execute, diagnose, and remediate them. We don't configure financial systems — we build agents that learn accounting rules and produce correct configurations. We don't coordinate cross-team handoffs — we build agents that do it for each other.
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
Key Job Responsibilities:
*Design and build autonomous AI agents that execute end-to-end ecommerce validation without human intervention
- Build orchestration systems that coordinate dozens of domain-specific agents working in parallel
- Create knowledge base architectures that ground agent behavior in accounting, tax, and compliance rules
- Solve novel problems at the intersection of AI, distributed systems, and financial domain complexity
A day in the life
An SDE II on Leo Regulus will be responsible for designing, building, and operating autonomous AI agent systems that automate the end-to-end ecommerce launch lifecycle — from requirements generation through financial configuration, test execution, and validation — enabling Amazon Leo to scale country and feature launches across 100+ markets without proportional headcount growth.
About the team
We are a team building the AI infrastructure that makes Amazon Leo's 100-country launch possible. Our entire stack is AI-native — every system we build either is an agent, powers an agent, or feeds an agent. We don't maintain legacy test frameworks or manual runbooks. We build autonomous systems and point them at problems that would otherwise require hundreds of engineers.
