As the founding engineer for Finance Automation inside Coinbase's FP&A & Treasury organization, you'll architect the AI-native platform that transforms how the company plans, allocates resources, and manages liquidity. Reporting to the VP of FP&A & Treasury, you'll set the technical direction for how FP&A scales planning, cost allocation, headcount systems, and treasury operations without scaling headcount. This is a forward-deployed engineering pilot where you own the make-vs-buy strategy and write the playbook for how AI transforms non-technical functions company-wide.

What you'll do:

• Own the design and delivery of AI agents that automate high-volume Finance workflows including period-end close, GL reconciliation, intercompany eliminations, and regulatory filing drafts across 60+ entities
• Architect a governed agent platform with observability, evaluation frameworks, grounding strategies, and human-in-the-loop controls that meet SOX audit requirements for every automated financial output
• Build integrations with core financial systems (NetSuite, Workday, Anaplan), designing data pipelines from sub-ledger through GL, consolidation, and multi-GAAP reporting
• Define compliance-grade engineering patterns where every agent action is traceable, reviewable, and defensible to auditors, with segregation of duties and evidence requirements as first-class concerns
• Partner with FP&A and Treasury leadership as the technical thought partner who translates ambiguous, spreadsheet-based workflows into engineered systems and communicates tradeoffs to non-technical stakeholders

Required Skills and Experience:

• 12+ years of backend software engineering with production systems at scale, including demonstrated experience building reliable distributed systems and idempotent document-processing pipelines
• Proven track record deploying LLM-based applications in production with evaluation frameworks, hallucination guardrails, and human-in-the-loop design for high-stakes outputs
• Multi-agent orchestration experience, including systems where agents collaborate, hand off tasks, and escalate to humans through state-managed approval workflows
• Financial systems literacy covering ERP data flows, reconciliation logic, trial balances, and the architectural constraints of SOX-controlled environments (does not need to be a CPA, but must be eager to learn)
• Demonstrated success operating as a solo or embedded engineer in a non-engineering org, setting technical direction autonomously and communicating effectively with non-technical domain experts
• Utilizes generative AI responsibly, maintaining human oversight to deliver business-ready outputs and drive measurable improvements in workflow efficiency, cost, and quality.

Job ID: P77680