Lead AI Architect - Universal Agent Control

EY; Hoboken (NJ); Hybrid; USD 144,000 - 329,000; Full time

Benefits offered by this job

Compensation package

Hybrid work model

Flexible vacation

Job summary

EY is seeking an experienced Lead AI Software Architect to lead a team of domain architect experts in designing AI Agent Control. The role covers tool usage, state management, policy definition and multiagent API development, with emphasis on authentication, observability, and confidential data handling.

You will mentor engineers, drive cross-functional collaboration, and shape a universal agentic control harness while contributing to IP filings and academic collaborations.

Qualifications

• Advanced degree in Electrical Engineering, Computer Science or Mathematics.
• 10+ years of experience in software engineering and AI.
• Experience with multi-agent system administration and modern secrets management.

Responsibilities

• Design control harness intelligence systems for multiagent environments.
• Design architectures for agent identity and authentication at scale.
• Create post-training and tuning environments for agents.
• Develop synthesis models for AI open-source models.
• Implement and benchmark architectures against agent and human evals.
• Build comprehensive AI Agent control frameworks.
• Lead AI Native design processes across teams.
• Research and assess emerging agent authorization architectures.
• Collaborate with AI, Security, Data and Infra teams for practical outcomes.
• Support strategic business development for emerging tech.

Skills

Analytical skills

Lead architecture

Mentor engineers

Cross-functional

Education

Advanced degree in STEM

Tools

Cursor

Claude Code

Qodo/Codium

Gemini

AWS Q

GitHub

NVIDIA NVAIE

Azure AI

Job description

EY is seeking an experienced Lead AI Software Architect to lead a team of domain architect experts in designing AI Agent Control. The role covers tool usage, state management, policy definition and multiagent API development, with emphasis on authentication, observability, and confidential data handling.

You will mentor engineers, drive cross-functional collaboration, and shape a universal agentic control harness while contributing to IP filings and academic collaborations.