We are
At Cross River, we're 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 - we’re reshaping how financial technology is developed and delivered. .
עיקרי התפקיד
As AI capabilities accelerate across the bank, we need an engineer to design and enforce safe AI usage—protecting customer data, preserving model integrity, and meeting our regulatory obligations. You'll be the architect of guardrails, tooling, and policies that make AI both secure and useful for product and internal teams. This isn't about slowing things down; it's about building the trust layer that lets innovation move fast without breaking things.
דרישות
You're a security engineer who's excited about the AI wave—someone who sees LLMs and Agentic AI as fascinating puzzles to secure, not just threats to mitigate. You've spent 5+ years in Security Engineering, AppSec, or Cloud Security, and at least 1–2 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 use—and where trust boundaries break down. You've 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.
You're 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 tooling—not 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 DSS—and actually having the evidence to prove it.
What You’ll Actually Be Doing
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, least‐privileged 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 third‑party 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 red‑teaming: automated jailbreak/prompt‑injection tests, safety benchmarks, purple‑team exercises, and response playbooks for AI incidents.
Enable the org with concise guidelines, examples, and training on safe AI development and usage.
5+ years in Security Engineering/AppSec/Cloud Security (or similar), including 1–2+ years securing AI/ML or data‑intensive systems (GenAI preferred).
Hands‑on 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/fine‑tune 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.
יתרון
Financial services background or other high‑assurance 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 red‑teaming or safety evaluation harnesses; contributions to OSS or published talks.
Why You’ll Love Working Here
Flexible hybrid work model: three days a week at our Jerusalem office
Monthly wellness reimbursement – from therapy to gel manicure, it's 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