Senior AI Engineer_CNTR

This is a Senior AI Engineer role owning a finance domain end-to-end — from context engineering and prompt design through tool use, evals, guardrails, and the UX around them — working directly with controllers, accountants, and CFOs to understand workflows and replace manual processes with production-grade agent systems. This is a backend-heavy, full-stack-in-practice role: no PM writes your specs and no architecture committee gates your ideas. The hardest problems on the roadmap involve building AI agents that finance teams and auditors can trust, giving agents the right financial context over large and messy data, and orchestrating durable workflows across flaky enterprise systems.

Details
Location: New York City, NY · In-person
Employment: Full-time
Experience: 2 years (agent-building), 4 years overall software engineering
Schedule: Startup hours, 6 days a week, 9am to 7pm or 8pm - IMPORTANT!
Visa & work authorization: None available

About the company
A seed-stage company building the AI operating system for the CFO office, with audit-ready AI agents that connect to a company's existing finance stack and automate the full order-to-cash, record-to-report, treasury, and financial-reporting workflow. The goal is for finance teams to review exceptions while AI handles the rest. Raised $9M from a well-regarded investor group that includes prominent AI and fintech leaders.

What you'll own
Own a finance domain end-to-end, building the agent system that automates it across context, prompts, tools, evals, guardrails, and UX, working directly with controllers and CFOs
Design and implement durable, replay-safe orchestration for long-running AI workflows across flaky, stateful enterprise systems
Build the trust layer for non-deterministic systems: evals, verification, guardrails, and observability that catch agent mistakes before a human does
Ship full-stack when the work calls for it, owning decisions across the LLM pipeline, infrastructure, backend, and UX within your pod
Do serious context engineering: retrieval, memory, and tool design that gets agents to reason reliably over large, messy, proprietary financial data
Build idempotent, audit-ready write-back systems for ERPs and financial systems with full traceability

What we're looking for
2 years specifically building and shipping AI agents for real-world, production problems; can point to something live and explain how you made it reliable
4 years of software engineering experience, mostly backend or infrastructure
Research grounding in AI agents, LLMs, or machine learning, ideally peer-reviewed work at top venues (NeurIPS, ICML, ICLR, ACL, or similar)
Strong backend proficiency in a modern language; Python is the primary stack
Worked at an early-stage startup (pre-seed through Series B)
Able to work in-person at the New York City office on the stated schedule (6 days a week, 9am-7/8pm)

Nice to have
Experience in fintech, ERP, accounting, payments, banking, treasury, audit, or compliance software
Shipped agents that take real action on production systems (money movement, ledger writes, system-of-record updates) with safety, idempotency, and rollback as first-class concerns
Built trust layers for non-deterministic systems: evals, verification, guardrails, and observability in production
Deep context engineering experience: retrieval, memory, and tool design over large, messy, proprietary data
Graduated post-2021, with direct exposure to the agentic AI wave from early in your career