Profile cover
J

AI Synthetic Red Team Attack Generator

Jackson Li

Full-time · Job · New_York

About the role

Manual red teaming scales poorly. A team of 10 red teamers working full-time can probe a model's guardrails with thousands of adversarial prompts per week. The model is responding to millions of real user queries in that same window. Synthetic attack generation — automatically producing diverse, novel adversarial prompts across the full taxonomy of potential violations — is what makes red teaming coverage comprehensive rather than sampled. We build LLM safety infrastructure for 55 enterprise clients deploying language models in regulated and high-stakes contexts. Synthetic red team attack generation is our evaluation backbone: the system that continuously generates novel adversarial test cases, tracks which attack categories penetrate current guardrails, and drives the prioritization for guardrail improvement. You'll design and build this system: attack taxonomy construction covering jailbreaks, injection, NSFW elicitation, PII extraction, and domain-specific violations for each client context, automated attack generation that produces semantically diverse variants within each category, and the evaluation harness that measures guardrail coverage against the generated attack distribution. The role is hybrid — 2 days per week in NYC for red team collaboration.

Responsibilities

  • Design attack taxonomy covering jailbreaks, prompt injection, elicitation, and domain-specific violation categories
  • Build automated attack generation producing semantically diverse variants within each taxonomy category
  • Implement evaluation harness measuring guardrail coverage rates against the synthetic attack distribution
  • Develop client-specific attack customization adapting taxonomy to regulated domain requirements
  • Create attack novelty tracking ensuring generated attacks don't just repeat known patterns
  • Build coverage gap reporting identifying which attack categories expose current guardrail weaknesses

Requirements

  • 6+ years in ML engineering with LLM safety, adversarial ML, or red teaming experience
  • Deep understanding of LLM failure modes and adversarial prompt construction
  • Strong Python and LLM API skills with prompt engineering expertise
  • Experience with NLP-based text generation and semantic diversity evaluation
  • Commitment to responsible security research and authorized safety evaluation contexts

Benefits

  • Hybrid: 2 days NYC, 3 days remote
  • Top-of-market compensation for LLM safety expertise
  • Comprehensive health benefits
  • $5,000 annual conference and research budget
  • Equity participation

Keep exploring related roles

Use these topic pages to discover more roles in the same hiring cluster.

Categories

NLP & Prompt Engineering AI & Machine Learning Engineers

Skills

python pytorch hugging face langchain fastapi

Related remote AI jobs

Nearby roles that share similar skills, categories, or hiring intent.

Browse all jobs

Senior AI Researcher- Pre-training (f/m/d)

RemoteAI company

Full-time Senior

Technical Writer

RemoteAI company

Part-time

AI Resident UK/EMEA

RemoteAI company

Full-time

Senior AI R&D Engineer – Model Training (f/m/d)

RemoteAI company

Full-time Senior

Senior Machine Learning Engineer (Clinical Team)

RemoteAI company

Full-time Senior

Conversational Modelling Research Engineer

RemoteAI company

Full-time

Job Type

Full-time

Level

Senior

Language

English

Salary Range

$160k – $200k / year

AI Expertise

NLP & Prompt Engineering AI & Machine Learning Engineers

Skills

python pytorch hugging face langchain fastapi sql

Ready to apply for this role?

Create a free talent account in under 2 minutes.

  • Apply to verified AI companies
  • Get AI-matched job recommendations
  • Message hiring managers directly
  • Build your public AI talent profile
Create free account & apply Log in

Interested in this role? Create your free profile to apply.

Apply — create free profile