AI Red Team Engineer

present of AI; Austin (TX); On-site; USD 120,000 - 160,000; Full time

Job summary

Hamming AI in Austin, TX seeks an engineer to build and run automated adversarial tests against voice and chat agents at scale, with permission. You will design evaluators and run large, concurrent campaigns to test AI agents and ensure fixes hold across releases.

You will work hands-on with Python/TypeScript backends, async workers, and failure handling, applying fuzzing or CTF techniques to surface meaningful failures. This role blends security testing with AI assurance.

Qualifications

• Candidate has backend or test infrastructure experience with Python or TypeScript.
• Comfort with async workers and failure handling is expected.
• Adversarial mindset from security tooling, fuzzing, or CTFs is required.

Responsibilities

• Build and run automated adversarial tests against voice and chat agents at scale.
• Turn each finding into a repeatable test and automation.
• Design evaluators to classify outcomes and guide fixes.

Skills

Adversarial testing

Security tooling

Fuzzing

CTF

Tools

Python

TypeScript

Job description

The job is to build and run the machinery that attacks voice and chat agents at scale, with the owner's permission, and to turn each finding into a repeatable test. A strong candidate brings solid backend or test infrastructure work in Python or TypeScript, comfort with async workers and failure handling, and a genuinely adversarial mindset from security tooling, fuzzing, or CTFs. Good judgment about evidence matters, knowing when an automated check is enough and when a person needs to look.

How it fits the AI era

The work sits directly on AI systems, probing how conversational agents can be tricked and making sure fixes hold across releases. Much of the repetition, variation, and replay can be automated, so the human part is designing the attacks, deciding what a result really means, and judging which failures matter. The role grows as more enterprises put agents on phone lines that anyone can reach.

How it builds AI experience

Someone here would gain hands on practice attacking LLM driven voice and chat agents, building evaluators that classify outcomes, and running large concurrent test campaigns against authorized targets. That experience comes from working the harness itself, learning how models and agent frameworks fail under pressure. Building a small fuzzing or red team project against a public agent would be a reasonable way to prepare.

Worth showing

• Python or TypeScript backend
• Async workers and queues
• Fuzzing or CTF work
• Reproducible evidence and replay

This is our own read, written from the employer's posting and the company's Spotlight. The employer's posting is the source of truth for the role.

About Hamming AI

Complete QA testing, red-teaming, and production observability platform for conversational voice AI agents, per the company.