Field AI Security Engineer - Enterprise Harness Specialist

Nullify; San Francisco (CA); On-site; USD 180,000 - 240,000; Full time

Job summary

Nullify is building a post-human product security harness that prioritizes high-impact work by continuously allocating AI capacity. You will ensure agents operate reliably in real enterprise environments and across diverse codebases, architectures, and security programs.

You’ll diagnose failure modes and edge cases to improve reasoning, context, tools, evaluation, orchestration, and the product itself. You’ll work with some of our largest customers, owning agent performance and extending the

Qualifications

• Strong fundamentals in Go and/or Python, and the judgment to know when to reach for which.
• Experience with production distributed systems—queues, state machines, idempotency, failure modes.
• A preference for simple, correct code over clever abstractions.
• Curiosity about security, even if it’s not your background yet.
• Customer obsession and passion for servicing customers and making them successful.

Responsibilities

• Own the performance of Nullify’s agents across a set of large enterprise customers.
• Find where agents and subagents fail, underperform or require human intervention—engineer those weaknesses out.
• Work against real customer codebases and environments to understand failure modes not shown in benchmarks.
• Build and improve agent tooling, context, orchestration, evaluation and underlying product capabilities.
• Turn customer-specific edge cases into generalized harness capabilities.
• Collaborate with customer security and engineering teams to define what good performance looks like in their environments.
• Feed field learnings into the core product to increase autonomous operation across more customers.

Skills

Go

Python

Distributed systems

Code quality

Security awareness

Job description

Nullify is building a post-human product security harness that prioritizes high-impact work by continuously allocating AI capacity. You will ensure agents operate reliably in real enterprise environments and across diverse codebases, architectures, and security programs.

You’ll diagnose failure modes and edge cases to improve reasoning, context, tools, evaluation, orchestration, and the product itself. You’ll work with some of our largest customers, owning agent performance and extending the