Builds production guardrails, abuse-detection systems, adversarial-testing infrastructure, and enforcement tooling to prevent model misuse across deployed AI products. Owns the safety layer between model capabilities and end users.
skills
[Python][adversarial ML and red teaming][abuse detection at scale][guardrails architecture (NeMo Guardrails, Garak)][large-scale data pipeline engineering][incident response for AI harms]
pay
low:320000
high:485000
currency:USD
basis:base
seniority
senior-ic
distinctFrom
Unlike ai-red-teamer (offensive probing to find vulnerabilities), this role owns the full defensive lifecycle: designing, building, shipping, and maintaining production safety systems and enforcement infrastructure. Unlike ai-governance-lead (policy), this is hands-on engineering.
Pay $320,000-$485K base confirmed across multiple boards (greenhouse.io, jobsbyculture, jobera). Frontier-lab (Anthropic) defensive safety engineering: production guardrails, abuse-detection infrastructure, and enforcement tooling. The London equivalent pays 240K-325K GBP.
