AI Security Engineer

As an AI Security Engineer , you will focus on securing cutting-edge Artificial Intelligence (AI) and Machine Learning (ML) systems within our enterprise banking platform. In this role, you will address security risks uniquely associated with AI infrastructure, Large Language Models (LLMs), and autonomous decisioning pipelines.

You will lead threat modeling, design automated detection systems, execute offensive security assessments, and build specialized response mechanisms to defend the bank from adversarial machine learning attacks (e.g., prompt injection, data poisoning, model extraction).

• Location: London, UK (Hybrid)
• Sector: Highly Regulated Financial Services / Banking
• Experience Level: Senior / Specialist

Primary Roles & Responsibilities

1. AI Threat Modeling & Risk Engineering

• Perform deep-dive Threat Modeling assessments on enterprise AI deployments and workflows to expose systemic architectural flaws before production deployment.
• Leverage AI Literacy to evaluate foundational model risks, focusing specifically on vulnerabilities mapped out by the OWASP Top 10 for LLMs and the MITRE ATLAS framework.
• Design, build, and scale automated security architectures ( Security Engineering ) that establish secure-by-design boundaries for multi-tenant AI integrations.

2. Offensive Security Testing & Detection Engineering

• Conduct automated and manual adversarial simulations ( Offensive Security Testing ) against deployment endpoints, evaluating model resilience against adversarial evasion, inversion, and data poisoning.
• Architect and implement specialized rules ( Detection Engineering ) to detect malicious queries, prompt injections, and data exfiltration patterns within streaming log structures.
• Execute active Cyber Threat Hunting cycles across core infrastructure to trace subtle anomalies indicating compromised training data pipeline elements or model drift manipulation.

3. Secure Development & Infrastructure Automation

• Write production-grade, highly optimized software tooling using Go (Golang) and Python to build security guardrails, custom wrappers, and proxy monitors for LLM applications.
• Embed security scanning, model validation, and drift analysis into modern infrastructure systems using DevOps methodologies and automated pipeline practices.
• Operate natively inside an ecosystem governed by Agile Methodology & Tools (e.g., Jira, Confluence) to cleanly track security initiatives alongside fast-moving development sprint tracks.

4. Response & Forensics Engineering

• Design and deploy real-time monitoring and defensive guardrails ( Advanced Threat Protection ) to isolate and mitigate malicious traffic targeting production AI endpoints.
• Develop incident playbooks ( Response Engineering ) specifically tailored to handle active exploitation of machine learning environments.
• Lead structural investigation procedures ( Digital Forensics ) following an anomaly or attack vector event, performing root-cause analysis on logs, system runtime snapshots, and compromised data artifacts.

Secondary Roles & Responsibilities

1. Customer Centricity & Stakeholder Collaboration

• Champion a mindset of Customer Centricity , ensuring that heavy security layers and model guardrails do not negatively disrupt the user experience or performance parameters for banking consumers.
• Act as an expert consultant for application teams, translating dense AI vulnerabilities into plain, actionable mitigation plans.

2. Continuous Learning & Research Contribution

• Dedicate time to Continuous Learning by tracking the volatile, fast-evolving AI threat landscape, incorporating new vulnerability research straight into the team’s defense playbooks.
• Evaluate emerging third-party security platforms, proxy tools, and open-source packages to modernize the bank’s security capabilities.

Technical Stack & Skills Required

• Languages: Advanced proficiency in Python (for ML tooling and scripting) and Go Programming Language (for building low-latency security proxies and microservices).
• AI Security Frameworks: Comprehensive understanding of MITRE ATLAS , OWASP LLM Top 10 , and standard evaluation tools (e.g., Garak, Inspect).
• DevOps Ecosystem: Container security frameworks ( Docker, Kubernetes ), CI/CD engineering pipelines, and Infrastructure-as-Code platforms ( Terraform ).
• Security & SIEM Tools: Experience developing custom detection logic inside tools like Splunk, Elastic Security, or Datadog.

Regulatory Context

All deployments must be engineered to securely satisfy rigorous UK and EU regulatory frameworks governing automated banking solutions, specifically:

• DORA (Digital Operational Resilience Act) parameters for continuous ICT risk testing.
• FCA Guidance on Artificial Intelligence regarding transparent governance, model validation, and consumer safety rules.