AI Security Engineer

We are

At Cross River were building the financial infrastructure that powers global innovation. With our cutting-edge suite of embedded payments cards and lending solutions we enable millions of businesses and consumers to transact seamlessly and securely.

With 900 employees worldwide and an R&D center of over 160 employees in Jerusalem - were reshaping how financial technology is developed and delivered. .

As AI capabilities accelerate across the bank we need an engineer to design and enforce safe AI usageprotecting customer data preserving model integrity and meeting our regulatory obligations. Youll be the architect of guardrails tooling and policies that make AI both secure and useful for product and internal teams. This isnt about slowing things down; its about building the trust layer that lets innovation move fast without breaking things.

Youre a security engineer whos excited about the AI wavesomeone who sees LLMs and Agentic AI as fascinating puzzles to secure not just threats to mitigate. Youve spent 5 years in Security Engineering AppSec or Cloud Security and at least 12 of those years have been spent getting your hands dirty with LLMs AI Agents and MCPs.

You understand how agentic frameworks (LangGraph CrewAI AutoGen and similar) orchestrate multi-step tool useand where trust boundaries break down. Youve assessed or secured AI-powered coding agents (Claude Code GitHub Copilot Cursor) and understand the unique risks of AI with filesystem terminal and API access in developer environments.

Youre equally comfortable dissecting a prompt injection attack as you are writing a Terraform module or shipping a Python library. You know your way around AWS and/or Azure modern app stacks (Python/TypeScript REST/gRPC containers/Kubernetes) and can translate security requirements into developer-friendly toolingnot just PDF policies that gather dust.

You communicate clearly in English and Hebrew thrive in regulated environments and understand that security in financial services means mapping controls to frameworks like FFIEC SOC 2 and PCI DSSand actually having the evidence to prove it.

• Design enterprise AI guardrails across Azure and AWS (e.g. Azure AI Studio/Azure OpenAI Amazon Bedrock/SageMaker): content filtering PII redaction prompt/response validation and policy enforcement services.
• Assess and define secure usage patterns and data governance controls for agentic frameworks and coding agents: permission scoping tool-call authorization leastprivileged retrieval session isolation and MCP server governance.
• Threat model AI systems (apps agents MCPs RAG fine-tuning pipelines) using frameworks like STRIDE and the OWASP Top 10 for LLM Apps; define misuse scenarios (prompt injection/context poisoning/jailbreaks/data exfiltration) and build mitigations.
• Build monitoring and telemetry: privacy-preserving prompt/response logging sensitive-data detection safety/eval dashboards drift/abuse signals and incident hooks into our SIEM.
• Integrate AI security into the SDLC: reusable libraries pre-commit checks CI/CD gates policy-as-code and secure-by-default reference architectures for product teams.
• Evaluate thirdparty AI vendors and internal apps: security reviews data residency and retention requirements SSO/SCIM integrations DPA/TPRM inputs and continuous control testing.
• Partner across Security Data Privacy and Engineering to map AI controls to FFIEC SOC 2 and PCI DSS; document control evidence for audits.
• Lead/participate in AI redteaming: automated jailbreak/promptinjection tests safety benchmarks purpleteam exercises and response playbooks for AI incidents.
• Enable the org with concise guidelines examples and training on safe AI development and usage.

• Flexible hybrid work model: three days a week at our Jerusalem office
• Monthly wellness reimbursement from therapy to gel manicure its up to you
• Full Keren Hishtalmut private health and dental insurance
• Volunteer days donation matching Yoga and Pilates
• A supportive collaborative culture that puts our people first

Hit Apply!

Requirements:

What You Bring to the Table

• 5 years in Security Engineering/AppSec/Cloud Security (or similar) including 12 years securing AI/ML or dataintensive systems (GenAI preferred).
• Handson experience with AWS and/or Azure and modern app stacks (Python/TypeScript REST/gRPC containers/Kubernetes IaC such as Terraform).
• Practical understanding of LLM attack surfaces (prompt injection context and goal poisoning data leakage via tools training/finetune poisoning model supply chain) and mitigation patterns.
• Experience assessing agentic architectures (LangGraph CrewAI or similar) and AI coding assistants (Claude Code GitHub Copilot) for secure enterprise deployment.
• Familiarity with identity and access for AI workloads (OAuth2/OIDC service principals role tokens PIM) and secure secret management/KMS.
• Experience implementing observability/telemetry and routing findings to SIEM; comfort balancing privacy with traceability.
• Ability to translate controls into developer-friendly libraries docs and CI/CD checks.
• Comfort working in a regulated environment and mapping controls to frameworks (FFIEC SOC 2 PCI DSS).
• Strong written communication in English and Hebrew.

Nice to have

• Financial services background or other highassurance domains.
• Exposure to Duende IdentityServer SSO/SCIM and enterprise authorization patterns.
• Familiarity with guardrail tooling (e.g. Azure AI Safety features Amazon Bedrock Guardrails) and policy engines (OPA/Rego).
• Prior work in AI redteaming or safety evaluation harnesses; contributions to OSS or published talks.