Cybersecurity AI Engineer III

Description

Cybersecurity AI Engineer III

Department: Information Security

Role Type: Engineering - Individual Contributor

Reports To: Director of Information Security

Level: Mid-level

The Cybersecurity AI Engineer III is a hands‑on engineering role responsible for building and operating secure AI‑enabled automations, agentic workflows, integrations, orchestration services, and reusable components within established cybersecurity and AI platform architectures. The role partners across Security Operations, IAM, Application and Cloud Security, Vulnerability Management, GRC, Data Security, Threat Intelligence, Enterprise Architecture, AI/ML platforms, software engineering, and model/technology risk. This role enables accelerated AI adoption while applying established engineering standards, trust boundaries, resilience requirements, and evidence expectations for a financial institution.

• Build, test, deploy, and operate bounded AI/agentic cybersecurity workflows for triage, enrichment, investigation support, control validation, evidence collection, remediation coordination, and analyst decision support using established architecture, design patterns, deterministic logic, workflow engines, LLMs, tool calling, RAG, and event‑driven orchestration as appropriate.
• Build and maintain reusable APIs, connectors, agent tools, plugins, event integrations, data transformations, and orchestration components for assigned use cases that securely connect approved models and agents to cybersecurity and developer platforms.
• Apply established machine-to-machine and agent-to-tool patterns using least privilege, workload identity, short‑lived credentials, scoped tokens, secrets protection, policy enforcement, network controls, encryption, versioned interfaces, comprehensive audit logging, and governed data handling/lineage.

• Embed AI‑assisted security into developer workflows, IDE/coding‑assistant ecosystems, source control, engineering portals, and CI/CD to support secure design, threat modeling, code/configuration review, dependency and vulnerability triage, policy‑as‑code, remediation guidance, test generation, evidence capture, and release‑risk summarization.
• Implement and maintain approved controls for AI coding agents and autonomous development workflows, including repository permissions, branch protection, code‑owner approvals, sandboxing, tool allow lists, secrets protection, artifact provenance, test gates, deployment authorization, and secure‑by‑default patterns for AI‑generated code, infrastructure‑as‑code, tests, and agent‑initiated changes.

• Implement and maintain approved guardrails for prompt/indirect prompt injection, unsafe tool use, excessive agency, sensitive‑data leakage, retrieval poisoning, insecure output handling, unauthorized cross‑system actions, and model/tool misuse. Apply established patterns for agent identity, authorization, trust boundaries, context/memory controls, secure retrieval, transaction‑level auditability, approval gates, segregation of duties, execution limits, rollback, fallback, and kill‑switch mechanisms;
  ** escalate
  * * new or material design decisions to senior engineers or security architects.
• Build and operate evaluation and red‑team harnesses for jailbreaks, adversarial inputs, unsafe actions, hallucinated evidence, privilege escalation, exfiltration, and control bypass; implement approved model/agent routing and policy rules based on data sensitivity, use‑case risk, geography, business context, and approved AI services.
• Instrument services with metrics, traces, logs, decision/tool‑call histories, latency, cost/token usage, quality/error measures, approval outcomes, policy violations, abnormal behavior, and integration health. Apply production engineering rigor including source control, automated testing, IaC, release pipelines, environment separation, SLOs, runbooks, rollback, and incident response.

• Support senior engineers, architects, cybersecurity risk, technology/model risk, privacy, legal, compliance, records…