Staff Applied AI Engineer

HartleyCo; New York (NY); On-site; USD 250,000 - 300,000; Full time

Job summary

HartleyCo in New York City is seeking an engineer to join an early-stage company building production AI agents for accounting. The role is five days onsite in NYC, with a base salary of $250,000–$300,000 plus meaningful equity.

You’ll own agent systems end to end: architecture, tool use, context, orchestration, evals, observability and reliability across real financial workflows. Staff-level engineering with strong Python fundamentals and production LLM/agent systems experience are essential.

Qualifications

• Staff-level engineering with strong Python fundamentals.
• Proven experience building production LLM or agent systems.
• Ownership of complex systems from prototype through deployment.
• Experience in early-stage, founding, or high-ownership startups.
• Strong judgement of deterministic code versus model-driven behaviour.
• Experience in regulated or document-heavy verticals (plus).

Responsibilities

• Own agent systems end to end: architecture, tool use, context, orchestration, evals, observability and reliability across real financial workflows.

Skills

Python

LLM/agent systems

End-to-end ownership

Job description

New York City | Five days onsite

$250,000 - $300,000 base + meaningful equity

We’re working with an early-stage company building production AI agents for accounting. They’ve raised $5m and are combining AI-native software with the acquisition of accounting firms - using technology to transform the underlying service rather than simply selling another software tool.

You’ll own agent systems end to end: architecture, tool use, context, orchestration, evals, observability and reliability across real financial workflows.

• Staff-level engineering ability with strong Python fundamentals
• Significant experience building production LLM or agent systems
• Evidence of owning complex systems from prototype through deployment
• Early-stage, founding or high-ownership startup experience
• Strong judgement around deterministic code versus model-driven behaviour
• Ideally, experience in a regulated or document-heavy vertical