Security engineer for AI agents

грейд
senior
формат
гибрид Лондон
язык
Английский B1
опубликована
сегодня

описание

This is a UK-based role; applicants must have the right to work in the UK.

We’re looking for a hands-on Security Engineer to help design, build and operate secure cloud platforms, data pipelines and data-sharing solutions. These capabilities support our internal product teams and deliver data to clients—either in client-owned cloud environments or in our own. The role is GCP-first, with security patterns that can be applied across AWS and Azure.

задачи

  • Build and improve secure cloud foundations in GCP, including resource hierarchies, organisation policies, VPC Service Controls, private connectivity, customer-managed encryption keys (CMEK) and centralised logging.
  • Develop reusable Terraform modules and project factories so teams can provision, test and redeploy secure environments consistently.
  • Apply policy-as-code in CI/CD, use keyless pipelines through workload identity federation, and automate detection and remediation of configuration drift.
  • Help establish equivalent security patterns in AWS and Azure.
  • Help design and evolve secure patterns for MCP hosts, clients, servers, gateways and registries.
  • Build or operate MCP gateways and registries with approved-tool allowlisting, tool-level authorisation, rate limits and auditable records of tool calls.
  • Implement OAuth 2.1/OIDC authorisation using short-lived, scoped and audience-bound tokens; avoid token passthrough and require human approval for high-impact actions.
  • Help secure MCP services hosted on Cloud Run or GKE, using controls such as Identity-Aware Proxy and prompt or response screening.
  • Threat-model and help defend against prompt injection, tool poisoning, unauthorised tool changes, confused-deputy attacks and data exfiltration through chained tool calls.
  • Help secure the agent and connector supply chain through vetting, signing, versioning, scanning and sandboxing.
  • Implement identity patterns for people, workloads and agents across Microsoft Entra ID and Okta, including federation, SSO, SCIM, Conditional Access and phishing-resistant MFA.
  • Extend identity governance to service accounts, API keys, workload identities and AI agents, using Saviynt or a comparable identity governance platform.
  • Support clear ownership, lifecycle management and regular access reviews for non-human identities.
  • Apply zero-standing-privilege principles through just-in-time elevation, short-lived credentials and automated access revocation.
  • Assess emerging capabilities such as Microsoft Entra Agent ID and Okta Cross App Access.
  • Build data transformation pipelines and governed data products using dbt, SQL and Terraform.
  • Apply controls such as data classification, lineage, model contracts, policy tags, masking, row-level security and least-privilege access.
  • Help deliver data securely to clients and partners through BigQuery Analytics Hub, cross-cloud storage transfers and API integrations.
  • Support secure ingestion, cleansing, enrichment and transformation, with scoped access, encryption, audit logging and monitoring at each hand-off.
  • Develop secure data-sharing patterns, including data clean rooms and consent-aware pipelines.
  • Help prevent sensitive data from leaking into agent prompts, context windows, logs or third-party models, and ensure agents access curated data rather than unrestricted raw tables.
  • Bring brand social accounts, advertising platforms and martech tools under central identity control using Cerby or a comparable solution.
  • Support secure onboarding and offboarding, credential rotation, enforced MFA and time-limited, auditable access for agencies, freelancers and creators.
  • Treat social content—including comments, direct messages, reviews and mentions—as untrusted input.
  • Design guardrails so agents cannot combine access to private data, untrusted content and the ability to publish externally without appropriate safeguards.

требования

  • Seasoned experience in cloud infrastructure, platform, data or security engineering, including hands-on GCP experience in production.
  • Strong practical GCP skills, such as IAM, organisation policies, networking, VPC Service Controls, Cloud Run or GKE, and BigQuery.
  • Working knowledge of comparable security capabilities in AWS and Azure.
  • Strong Terraform experience, including reusable modules, environment and state management, testing, CI/CD and policy-as-code.
  • Experience engineering identity controls with Microsoft Entra ID and/or Okta, including SSO, SCIM, Conditional Access, MFA and privileged or just-in-time access. Familiarity with Saviynt or another IGA platform is expected.
  • Hands-on SIEM and SOAR experience, including onboarding log sources, developing detections and building response playbooks.
  • Practical understanding of MCP and agent security. This could come from production work, open-source contributions or well-developed proofs of concept.
  • Experience with dbt and SQL for data transformation or data products, and a working understanding of data-layer controls such as classification, masking, row-level security and access grants.
  • Experience delivering data to clients or partners across organisational or cloud boundaries, using services such as BigQuery Analytics Hub, GCS, S3, Azure Blob Storage or APIs.
  • Coding ability in Python, Go or TypeScript for automation, tooling or MCP server development.
  • Familiarity with social, advertising or marketing platforms and the security challenges involved in managing brand accounts and non-SSO applications.
  • You build securely by default and consider how systems could be misused from the outset.
  • You automate repeatable work, test what you codify and look for ways to make secure delivery easier.
  • You balance risk, usability and delivery needs, and can explain when a solution is not ready—and what would make it safe to proceed.
  • You communicate clearly with technical and non-technical colleagues and can explain security risks in plain language.
  • You document decisions and create practical standards, golden paths and runbooks.
  • You measure outcomes, learn from incidents and feedback, and improve controls over time.
  • You collaborate inclusively, seek different perspectives and support the development of colleagues.
  • You’ll be familiar with relevant security, privacy and AI frameworks, such as ISO 27001, SOC 2, NIST CSF 2.0, NIST AI RMF, ISO/IEC 42001, CIS Benchmarks, UK GDPR/GDPR and the EU AI Act.
  • You should hold—or be willing to achieve after joining: Google Cloud Professional Cloud Architect within six months.
  • You should hold—or be willing to achieve after joining: Google Cloud Professional Cloud Security Engineer within 12 months.
  • Будет плюсом: Experience operating MCP servers or gateways in production, or contributing to open-source agent or security tooling; Google Security Operations or Microsoft Sentinel experience, particularly detection-as-code practices; Kubernetes or service-mesh security experience, including workload identity, policy controllers or mTLS; Software supply-chain security experience, such as SLSA, Sigstore/cosign, SBOMs or Binary Authorization; Privacy engineering or AI assurance experience, including data clean rooms, consent management, pseudonymisation or AI red-teaming; Familiarity with MITRE ATLAS, OWASP GenAI guidance or AI security testing; Agency, media or consumer-brand experience, particularly in environments requiring multi-client data segregation; Further relevant certifications, such as Google Professional Security Operations Engineer, Google Professional Cloud DevOps Engineer, HashiCorp Terraform Associate, Microsoft SC-300/AZ-500, Okta Certified Professional, AWS Certified Security – Specialty or CCSP

условия

  • WPP will fund exam fees and provide dedicated study time.
  • Our work is powered by a commitment to Be Extraordinary: investing in our people and supporting transformational creativity. We Lead Optimistically, champion growth and development, and create opportunities for colleagues to build connections and learn across WPP Media and WPP’s global network.
  • We aim to create an open environment built on trust and respect, where everyone feels they belong and can progress. Employee resource groups, in-office events and opportunities to share ideas help bring our teams together. WPP Media also offers a comprehensive benefits package; specific benefits vary by location.
  • Teams typically work in the office around four days a week to support collaboration and connection. If you need an accommodation or flexibility, please discuss this with the hiring team.

навыки

компания

похожие вакансии

статьи для DevOps-инженеров