Director, AI ML and Cloud Security

Description

Director, AI ML and Cloud Security: a hands-on engineering leadership role within the Chief Security Office.

The company is a global biopharma business working to get ahead of disease. This role leads AI engineering delivery for the Cyber Architecture and Solutions team, which builds what the AI architecture group designs and then passes it to a managed service vendor to run.

The AI ML and Cloud Security function is changing how security operates across the Chief Security Office. The team is building an agentic framework, a multi-agent platform that moves security services from rule-based to AI-driven operation. It owns AI architecture and engineering end to end: the agent layer, LLM orchestration, the AI and ML platform and the data fabric pipeline. It also governs the security posture of AI tools, advises on cloud security across the Azure and GCP estate, and connects business units and R&D with the technical AI function. The AI security standards it sets must be followed by every AI initiative in the company. There are 17 roles across the UK, India and the US.

You will work hands on: reviewing code, debating orchestration patterns and judging whether a build is truly ready for a vendor to run at 3am. Your engineers work in LangChain, LlamaIndex, vector databases and production Python on Kubernetes, so you need to read that work critically. Others set the architecture. You may disagree with it, so challenge it properly, then build it well.

You are accountable for build quality, test coverage and handover readiness for vendor operation. You do not define the architecture.

Responsibilities

• Take ownership of AI engineering delivery across the agentic framework build, AI solution engineering, platform infrastructure and the data fabric pipeline.
• Build AI-powered security capability, such as GenAI alert triage, automated threat investigation and LLM-assisted analysis, and prove each one works before anyone relies on it.
• Set the engineering bar for the 4 workstreams, including test coverage thresholds, evaluation harness pass rates and the definition of handover-ready, and audit against it at every release.
• Chair the fortnightly build review with the managed service vendor, and sign or refuse the handover certificate for each agent and pipeline they take on.
• Hold the 4 workstream leads to one release train and settle conflicts when agent and platform work compete for the same engineer.
• Make sure every build is tested, documented and packaged so a managed service can run it without Cyber Architecture and Solutions staff present.
• Lead 6 engineers across India and the UK, and manage dependencies between the agent, platform and delivery workstreams.
• Build security controls in by design to the binding AI security standards, and secure sign-off before release.
• Act as the senior engineering voice in architecture reviews, testing designs for buildability, testability and operability.

Requirements

• Significant experience in cyber security or AI engineering, including leading technical teams across several geographies. Your domain focus is AI engineering, ideally having led an engineering team that delivered production AI at enterprise scale.
• Experience leading 4 engineering workstreams: an agentic framework build with a vendor, AI solution engineering (GenAI alert triage, automated investigation, LLM-powered tooling), a data fabric pipeline, and AI and ML platform engineering covering Kubernetes, CI/CD and resilience.
• Working knowledge of AI delivery standards: evaluation harnesses, red-team and prompt-injection testing, model and prompt versioning, observability, and cost and latency budgets.
• Ability to apply binding AI security standards in build, including tool-invocation control, identity binding, least privilege, guardrails, output filtering and full audit traceability.
• Experience building to an architecture specification and signing off vendor handover, delivering work that is built, tested, documented and ready to run as a managed service. You raise design problems back rather than engineering around them.
• Skills tagged on the listing include AI for cybersecurity, application security, CI/CD, cloud computing, cloud security, computer programming, data security, Google Cloud Platform, identity and access management, information systems, infrastructure security, Kubernetes, LangChain, LangGraph, large language models, LlamaIndex, MLOps, Microsoft Azure, MITRE ATT&CK, MLflow, ML security, operating systems security, OWASP Top 10, Pinecone and 6 more.
• Preferred: an Azure security certification such as AZ-500, or Google Professional Cloud Security Engineer.
• Preferred: a working command of NIST AI RMF, ISO/IEC 42001, the EU AI Act and the OWASP Top 10 for LLM Applications, including where each one stops being useful.
• Preferred: familiarity with MITRE ATLAS and published adversarial machine learning research.
• Preferred: hands-on experience with LangChain or LlamaIndex, vector databases, production Python and Kubernetes.
• Preferred: experience evaluating and testing non-deterministic agent behaviour.
• Preferred: experience with AI security controls such as guardrails, tool-invocation control and least-privilege scoping.
• Preferred: hands-on MLOps or LLMOps practice, including model registry, prompt management, evaluation and cost monitoring in production.
• Preferred: experience coordinating delivery with architecture and AI security owners without redefining their design or policy.
• Preferred: Certified Kubernetes Security Specialist, or equivalent proven container security depth.
• Preferred: experience in GxP-regulated or other safety-critical settings where an automated action has real-world consequences.
• Preferred: CISSP or CCSP.
• Preferred: experience in pharmaceutical, life sciences or another highly regulated industry.

Conditions

Based in the United Kingdom on a hybrid model, with 2 to 3 days a week onsite at the central London headquarters.

Flexible working is encouraged, so ask the hiring team which options are available.

Applications close on 6 November 2026.

US licensed healthcare professionals may have interview-related expenses reported under US transparency rules.