AI Cybersecurity Engineer

Job Description AI Cybersecurity Engineer Who We Are

At Upbound Group, we are committed to elevating financial opportunity for all through innovative, inclusive, and technology-driven financial solutions that address the evolving needs and aspirations of consumers. The Company’s customer-facing operating units include industry-leading brands such as Rent‑A‑Center, Acima and Brigit that facilitate consumer transactions across a wide range of store-based and digital retail channels, including over 2,400 company‑branded retail units across the United States, Mexico, New York and Puerto Rico.

Upbound Group, Inc. is headquartered in Plano, Texas.

We are seeking a forward‑thinking AI Cybersecurity Engineer to join our Security team. This role sits at the convergence of Zero Trust architecture, Generative AI, agentic systems, and modern security engineering. The AI Cybersecurity Engineer will design, build, and operationalize next‑generation AI‑driven security capabilities - including autonomous security agents, Retrieval‑Augmented Generation (RAG) pipelines, and Model Context Protocol (MCP) integrated tool chains - to protect our infrastructure, data, and users against an ever‑evolving threat landscape.

This role is critical to enabling safe, responsible AI adoption across our brands while maintaining the trust of the consumers we serve.

• Apply Zero Trust principles to AI agents, ensuring agents operate under strict least‑privilege policies with scoped, time‑limited credentials.
• Secure GenAI deployments including LLM APIs, fine‑tuned models, and foundation model integrations against threats such as prompt injection, jail breaking, training data poisoning, and model inversion attacks.
• Build and maintain guardrails, content moderation layers, and output validation pipelines for GenAI systems and LLMs used in security and business workflows.
• Conduct adversarial red‑teaming of GenAI systems, agent platforms, and LLMs to identify exploitable behaviors, unsafe outputs, and data exfiltration risks; develop remediation playbooks.
• Secure multi‑agent systems (MAS) that autonomously perform security tasks such as threat hunting, incident triage, vulnerability scanning, and policy enforcement.
• Define agent trust boundaries, inter‑agent communication security, and human‑in‑the‑loop (HITL) checkpoints to prevent runaway or adversarially hijacked agent behavior.
• Implement agent observability frameworks - logging, tracing, and auditing all agent decisions, tool calls, and external API interactions for forensic and compliance purposes.
• Assess and mitigate agentic‑specific attack surfaces including goal hijacking, tool misuse, privilege escalation via chained tool calls, and unintended data exfiltration.
• Evaluate, harden, and govern the use of Model Context Protocol (MCP) servers that expose enterprise tools and data to AI agents - treating each MCP server as a security boundary requiring authentication, authorization, and audit logging.
• Define and enforce MCP server access control policies, ensuring agents can only invoke permitted tools within approved scopes and that all MCP tool calls are logged and attributable.
• Assess MCP‑specific risks including prompt‑injected tool invocation, unauthorized resource access through MCP resource endpoints, and lateral movement via chained MCP server calls.
• Collaborate with platform and integration teams to establish secure MCP deployment standards, including mTLS for server communication, secrets management for server credentials, and rate limiting for tool invocations.
• Harden RAG pipelines against retrieval manipulation attacks, indirect prompt injection via poisoned knowledge base documents, and sensitive data leakage through retrieved context.
• Design RAG pipeline monitoring and anomaly detection to identify unusual retrieval patterns, high‑entropy queries indicative of extraction attacks, and drift in retrieved context quality.
• Build and deploy ML models for real‑time threat detection, behavioral anomaly detection, and user/entity behavior analytics (UEBA) across network, endpoint, and cloud telemetry.
• Develop LLM‑powered SOAR integrations that automate alert…