AI Security Engineer

The world of digital assets is accelerating in speed, magnitude, and complexity, opening the door to new ways for leveraging the blockchain. Fireblocks’ platform and network provide the simplest and most secure way for companies to work with digital assets and it trusted by some of the largest financial institutions, banks, globally-recognized brands, and Web3 companies in the world, including BNY Mellon, BNP Paribas, ANZ Bank, Revolut, and thousands more.

About The Team

As Fireblocks leads AI adoption across development and business functions, we are seeking an AI Security Lead to drive our strategy and implementation of AI/ML security across the enterprise.

This role combines advanced engineering, adversarial risk defense, and strategic security oversight. You’ll serve as the company’s subject-matter expert on AI security risks, threat detection and mitigation, secure development practices, and AI-related compliance.

Your mission will include safeguarding against emerging threats in AI workflows, such as adversarial AI, AI-based malware, LLM prompt injection, and model tampering

If you’re the kind of person who sees where AI security is headed and wants to build the defenses before the threats hit, Fireblocks is the place. You’ll help define what secure-by-design really means in the age of AI, working at the intersection of deep engineering, cybersecurity, and innovation.

What You’ll Do

• Design and implement best-in-class security measures for the entire AI/ML lifecycle – from data pipelines and model training to deployment and inference.
• Address risks like model theft, data leakage, prompt injection, adversarial attacks, model inversion, deepfake phishing, and misuse of LLMs or vector databases.
• Lead threat modeling activities across AI/ML pipelines, RAG platforms, and use cases, and contribute to broader risk assessments and incident response strategies.
• Partner with R&D, MLOps, DevSecOps, and SecOps to ensure secure and ethical practices in AI usage, procurement, and model governance. Work with Red Teams to simulate AI-enabled attacks and develop countermeasures.
• Develop protections for internal AI agents and LangChain/RAG/AutoGen-based systems. Build and maintain libraries for prompt sanitization, input/output policy enforcement, and adversarial defense. Collaborate with developers to create layered prompt chains resilient to injection and context confusion.
• Integrate anomaly detection for prompt payloads, fine-tune poisoning, agent hallucinations, and monitor for AI-specific threats across environments.
  

• 5+ years in cybersecurity or application security roles
• Deep knowledge of AI security risks, including prompt injections, model theft, adversarial attacks, data leakage, and LLM abuse patterns.
• Experience with securing AI agent frameworks (e.g., AutoGen, LangChain, CrewAI) or AI-native apps.
• Familiarity with generative AI tools (OpenAI, Claude, Hugging Face, etc.) and integration patterns in cloud/SaaS platforms. Python proficiency with the ability to contribute secure AI-related tooling. Familiar with threats like WormGPT, FraudGPT, or BlackMamba.
  

• Experience with privacy-preserving ML (differential privacy, federated learning)
• Experience with MLOps tools: Kubeflow, SageMaker, MLflow, etc.
• Cybersecurity certifications (e.g., CISSP, OSCP) or ML credentials