Adversarial AI Engineer

GoFundMe.org

full-time

Posted on: 9/27/2025

Location Type: Hybrid

Location: San Francisco • California • 🇺🇸 United States

✨ AI Apply

💰 $181,000 - $271,000 per year

Mid-LevelSenior

AWSAzureCloudCyber SecurityGoogle Cloud PlatformPythonPyTorchTensorflow

About the role

Execute adversarial testing of LLMs, agentic AI systems, recommendation models, and fraud detection tools using techniques such as prompt injection, jailbreaking, data poisoning, model inversion, and membership inference attacks
Develop synthetic attack datasets tailored to fundraising and trust & safety scenarios
Build automated adversarial testing pipelines integrated into CI/CD and reusable robustness evaluation libraries
Deploy real-time detection for prompt injection and model evasion; implement input validation, output filtering, adversarial training, and differential privacy mechanisms
Lead red team operations and strengthen AI/ML systems powering fraud detection, content moderation, and Trust & Safety at scale
Partner with Trust & Safety, Product Security, and Data Science to mitigate algorithmic bias and implement security controls
Establish AI security policies, training, deployment review processes aligned with NIST AI RMF
Build monitoring and incident response systems for AI security
Stay current with emerging attack vectors; contribute to open-source adversarial tools and publish externally to advance GoFundMe’s AI security leadership

6–8 years in cybersecurity with a focus on AI/ML security or adversarial ML
2+ years specialized LLM security experience (prompt injection, jailbreaking, adversarial prompt crafting)
Proven red team / penetration testing background on AI systems
Strong Python programming with ML frameworks (TensorFlow, PyTorch, Hugging Face)
Deep understanding of ML fundamentals, Neural Networks, transformers (GPT, LLaMA, Claude, BERT) and known vulnerabilities
Experience testing Agentic AI security including agent frameworks (LangGraph, AutoGen, CrewAI, Google ADK, Pydantic AI)
Skilled in adversarial attack methods: data poisoning, model evasion, membership inference, model extraction
Knowledge of defense mechanisms: adversarial training, input sanitization, differential privacy, robustness certification
Hands-on adversarial attack experience: data poisoning, model evasion, membership inference, model extraction
Familiarity with OWASP Top 10 for LLMs, MITRE ATLAS, NIST AI RMF
Experience with threat modeling, security architecture, and cloud controls (AWS, GCP, Azure)
Candidates located in the San Francisco, Bay Area with an in-office requirement of 3x a week

Benefits

Tip: use these terms in your resume and cover letter to boost ATS matches.

adversarial testingprompt injectionjailbreakingdata poisoningmodel inversionmembership inferencePythonTensorFlowPyTorchadversarial training

leadershipcollaborationcommunicationproblem-solvingcritical thinking