We are AI-native by default, but AI is a means, not the goal. The bar is knowing where a model can genuinely improve an outcome, where deterministic systems must remain in control, and how to build products that are reliable, auditable, and trusted when money is involved.

What you'll be doing: - Own a vertical end-to-end — define outcomes and metrics, maintain a backlog prioritized by value, risk, dependency, and strategic priority, and set acceptance criteria crisp enough to build and test. - Shape the technical design with engineering — system boundaries, money movement correctness, third-party rail reliance, rail-down behavior, failure/retry/refund paths, security, latency, scalability, recovery. - Make hard trade-offs across correctness, speed, cost, reliability and UX, and defend them to engineering, compliance, or operations leads. - Treat regulated constraints as first-class product inputs — licensing scope, CDD/KYC and limits, sanctions/PEP screening, transaction monitoring, data residency and privacy. Identify hard launch gates vs forward-readiness. - Apply AI where it earns its place — define what it can decide, human in the lo

We are AI-native by default, but AI is a means, not the goal. The bar is knowing where a model can genuinely improve an outcome, where deterministic systems must remain in control, and how to build products that are reliable, auditable, and trusted when money is involved.

What you'll be doing: - Own a vertical end-to-end — define outcomes and metrics, maintain a backlog prioritized by value, risk, dependency, and strategic priority, and set acceptance criteria crisp enough to build and test. - Shape the technical design with engineering — system boundaries, money movement correctness, third-party rail reliance, rail-down behavior, failure/retry/refund paths, security, latency, scalability, recovery. - Make hard trade-offs across correctness, speed, cost, reliability and UX, and defend them to engineering, compliance, or operations leads. - Treat regulated constraints as first-class product inputs — licensing scope, CDD/KYC and limits, sanctions/PEP screening, transaction monitoring, data residency and privacy. Identify hard launch gates vs forward-readiness. - Apply AI where it earns its place — define what it can decide, human in the lo