Head of AI Agent Engineering

EY; Phoenix (AZ); Hybrid; USD 126,000 - 230,000; Full time

Job summary

EY is hiring to build AI agents that execute complex workflows from intent to verified outcomes. The role blends applied AI, distributed systems, and product engineering, delivering durable agent behavior and safe autonomy in production environments.

You will tackle long-running executions, enterprise integrations, and production failures, balancing determinism with value while maintaining user control and observability.

Qualifications

• Bachelor’s degree required or equivalent.
• 7–9 years of software or applied ML engineering experience.
• Hands-on experience shipping large-language-model applications or agent workflows.
• Proficiency in Python or TypeScript for production services.
• Experience with model APIs, structured outputs, and tool use.
• Strong knowledge of distributed/system design and failure recovery.
• Ability to design permissioned, bounded authority systems.
• Willingness to travel up to 50% and work in a hybrid capacity.

Responsibilities

• Choose the right execution model for deterministic vs. autonomous workflows.
• Build dependable tools with clear contracts and error handling.
• Engineer agent behavior across planning, routing, and handoffs.
• Create durable context with memory and environment feedback.
• Bound agent authority with scoped identities and budgets.
• Evaluate complete trajectories with tests and verified outcomes.
• Operate long-running execution with durable state and safe retries.
• Make behavior observable with end-to-end traces and replay.

Skills

Production engineering

Agent architecture

AI application depth

Evaluation discipline

Security and operability

Product judgment

Education

Bachelor’s Degree in a relevant field

Tools

Python

TypeScript

Job description

EY is hiring to build AI agents that execute complex workflows from intent to verified outcomes. The role blends applied AI, distributed systems, and product engineering, delivering durable agent behavior and safe autonomy in production environments.

You will tackle long-running executions, enterprise integrations, and production failures, balancing determinism with value while maintaining user control and observability.