Principal Software Architect, Finance Technology Platform

Principal Software Architect, Finance Technology Platform

We are seeking a Software Architect to define and lead the technical vision for Apple's next-generation Finance Technology Platform. This is a platform-defining technical authority role, responsible for designing and building a multi-tenant, AI-native platform that powers financial transaction processing end-to-end — from business event ingestion through validation, enrichment, approval, position tracking, settlement, and reconciliation. This is not a management role.

It is a first-principles engineering leadership role: you will architect the platform, set the standards other teams build against, and personally shape the agentic Finance AI foundations that make this platform fundamentally different from the transaction-recording systems that came before it.

Finance systems have historically been transaction recorders. We intend to build a platform that reasons. As Principal Software Architect, you will architect a platform where AI agents perform the majority of financial processing work end-to-end — interpreting business intent, validating and enriching transactions, routing approvals, matching and adjudicating documents, resolving exceptions, corresponding with counter parties, and preparing settlement — with humans in the loop by policy, not by default.

You will design the platform as a multi-tenant substrate: a shared canonical data model, orchestration engine, sub-ledger, accounting determination service, data layer, and reconciliation engine that multiple engineering teams build their own finance applications on top of, configuring behavior rather than rebuilding shared capability. Domain teams should onboard a new financial flow in weeks through configuration — not months through development.

You will influence architecture before code is written, own the standards that govern financial correctness and auditability, and be the senior technical voice partnering with Finance, Controllership, Tax, and Internal Audit leadership.

• Define and govern the end-to-end architecture for the Finance Technology Platform — transaction ingestion, validation, enrichment, matching, approval, position tracking, accounting determination, settlement, and reconciliation — at enterprise scale.
• Architect the platform as a multi-tenant, multi-entity foundation that hosts applications built by multiple engineering teams: tenant isolation, shared-capability reuse, per-tenant policy and configuration, versioned APIs, and independent release cadence — with new financial flows onboarding in weeks via configuration, not months of development.
• Own the agentic architecture of the platform: define how AI agents are specified, composed, evaluated, sandboxed, and promoted to production across transaction intake, approval reasoning, document extraction, match adjudication, exception resolution, counter party correspondence, and settlement optimization.
• Design the agent harness and its supporting infrastructure — tool/function interfaces over the platform's own APIs, graph-based orchestration and multi-agent hand-off, context assembly and short/long-term memory, retrieval over policy, contract, and master-data corpora, structured output validation, deterministic replay, and cost/latency budgets with graceful degradation to deterministic paths.
• Establish the evaluation and safety discipline for agentic finance: golden datasets, offline and online evals gating CI, confidence thresholds and human-in-the-loop escalation policy, guardrails against injection via externally supplied documents, full audit trails for agent actions, and clear boundaries where AI is never authoritative — money movement, posting, and control points stay deterministic and verifiable.
• Set the core platform patterns: one canonical document model, saga-based orchestration, a sub-ledger that keeps financial position separate from accounting determination, and business rules as versioned configuration rather than compiled logic.
• Own financial correctness and SOX controls: immutable event history, exactly-once settlement, full lineage from source document to GL, continuous reconciliation with clear break…