We are hiring a Lead AI Security Engineer to secure enterprise AI platforms, GenAI applications, LLM-based solutions, AI agents, and AI-enabled services across their full lifecycle. The role focuses on AI security engineering, threat modelling, architecture assurance, secure AI control design, adversarial testing, and practical risk mitigation within a complex global environment.
Responsibilities
Design and validate security controls for AI platforms, GenAI applications, LLM integrations, RAG pipelines, AI agents, model endpoints, and AI-enabled workflows.
Partner with engineering, platform, cloud, data, application security, and architecture teams to embed AI security requirements into design, build, deployment, and operations.
Identify and assess AI-specific threats including prompt injection, jailbreaks, insecure tool use, data leakage, model extraction, poisoning, RAG abuse, agent misuse, and AI supply chain exposure.
Conduct AI threat modelling, adversarial testing, attack surface reviews, control gap analysis, residual risk documentation, and mitigation planning across platforms such as DATALABS, AI Foundry, Azure Cognitive Services, Databricks, and Snowflake.
Define and review controls for model access, prompt and response handling, sensitive data protection, identity and access, secure APIs, logging, monitoring, abuse detection, and incident readiness for AI systems.
Use frameworks such as OWASP Top 10 for LLM Applications, MITRE ATLAS, NIST AI RMF, Responsible AI, and EU AI Act expectations as supporting references for control alignment and audit readiness.
Support AI cybersecurity standards, secure design patterns, assessment checklists, governance inputs, and awareness material.
Improve AI security metrics, risk visibility, control automation, and reusable security guidance.
Required qualifications
10 years of experience in cybersecurity, security architecture, application security, cloud security, AI/ML security, or enterprise security engineering.
Strong hands-on experience securing AI, ML, GenAI, LLM applications, RAG pipelines, AI agents, model APIs, or enterprise AI platforms.
Deep understanding of prompt injection, jailbreaks, insecure plugins/tools, model poisoning, data leakage, model exfiltration, RAG security, agent security, AI supply chain risk, AI red teaming, and misuse scenarios.
Working knowledge of OWASP LLM Top 10, MITRE ATLAS, NIST AI RMF, Responsible AI, and AI regulatory expectations.
Experience designing, validating, and operationalizing AI security controls, including secure architecture reviews, threat modelling, adversarial testing, remediation tracking, and risk acceptance support.
Strong knowledge of cloud security, application security, API security, data protection, IAM, logging, monitoring, security operations, and enterprise risk management as applied to AI systems.
Ability to work with engineering and senior stakeholders to explain AI security risks and convert them into practical technical controls.
Preferred qualifications
Experience with AI security engineering, AI red teaming, adversarial testing, model robustness evaluation, secure RAG design, agent security, or AI platform security in regulated environments.
Experience with enterprise AI platforms such as Azure OpenAI, Microsoft AI Foundry, Azure AI Services, Databricks, Snowflake, Microsoft Purview, or Microsoft security technologies.
Exposure to Responsible AI, model risk management, AI governance programs, Python scripting, automation, or relevant security certifications such as CISSP, CISA, CRISC, GIAC, CCSP, or ISO 27001.
For full details about this role, click “Apply on company site” to view the employer’s original posting.
Skills for this role
AI SecurityCybersecuritySecurity ArchitectureApplication SecurityCloud SecurityThreat ModellingAdversarial TestingGenAILLMRAGAI AgentsOWASP LLM Top 10MITRE ATLASNIST AI RMFPythonAutomationAzure OpenAIDatabricksSnowflake
Your skill match
Checking your profile…
YOUR NEXT STEP
Get interview-ready for this role
A focused preparation guide, built around this job’s responsibilities and requirements.
✦ AI-generated guide
Preparation suggestions, not the employer’s actual interview questions. Always check the original posting for current requirements.
01
Understand the role
What matters most
Hands-on AI Security Engineering
The role prioritizes practical execution over theoretical governance. You must demonstrate the ability to design and validate security controls for complex AI ecosystems, including RAG pipelines, LLM integrations, and AI agents, ensuring they are integrated directly into the development lifecycle.
AI Threat Modelling and Adversarial Testing
You are expected to lead the identification of AI-specific attack vectors such as prompt injection, model poisoning, and data exfiltration. This requires conducting rigorous adversarial testing and threat modelling across enterprise platforms like Azure AI, Databricks, and Snowflake.
Cross-Functional Stakeholder Enablement
Success depends on your ability to translate complex AI security risks into actionable technical requirements for engineering, data, and architecture teams. You must balance security rigor with the need for scalable, enterprise-ready AI delivery.
Framework Alignment and Compliance
You must operationalize security standards by mapping controls to industry frameworks like OWASP LLM Top 10, MITRE ATLAS, and NIST AI RMF, ensuring that AI deployments remain audit-ready and compliant with evolving regulations like the EU AI Act.
Likely challenges to think through
Balancing Security with AI Velocity
The primary challenge is implementing robust security controls without hindering the rapid development cycles of GenAI applications. Prepare by discussing how you automate security guardrails and integrate them into CI/CD pipelines to maintain speed while ensuring safety.
Securing Non-Deterministic AI Systems
Unlike traditional software, AI models exhibit non-deterministic behavior, making traditional security testing insufficient. Prepare to discuss how you approach adversarial testing and robustness evaluation for systems where inputs and outputs are highly variable.
Managing Complex Global Ecosystems
Operating within a large, regulated global environment requires navigating diverse technical stacks and organizational silos. Prepare to explain how you maintain consistent security standards across disparate platforms like Azure, Databricks, and Snowflake while managing residual risk documentation.
03
Practice your answers
Technical practice
Securing RAG Pipelines
What to coverExplain how you would design security controls for a RAG pipeline to prevent data leakage and unauthorized access to sensitive documents. Focus on the trade-offs between data retrieval accuracy and the implementation of access control lists (ACLs) at the vector database level.
Mitigating Prompt Injection
What to coverDescribe your approach to defending against indirect and direct prompt injection attacks in LLM-based applications. Discuss the effectiveness of input sanitization, output filtering, and the use of system-level guardrails versus model-based detection.
AI Threat Modelling
What to coverWalk through a threat modelling exercise for an AI agent integrated with external tools. How do you identify risks related to insecure tool use and unauthorized API calls, and what specific controls would you implement to mitigate these?
Adversarial Testing Strategies
What to coverHow do you structure an adversarial testing or red teaming program for GenAI models? Explain how you prioritize test cases based on the OWASP Top 10 for LLM Applications and how you document residual risk for stakeholders.
AI Supply Chain Security
What to coverWhat are the primary security risks associated with the AI supply chain, such as using third-party pre-trained models or open-source libraries? Suggest methods for validating model integrity and managing dependencies in a secure deployment lifecycle.
Monitoring and Incident Readiness
What to coverDefine the key metrics and logging requirements for detecting abuse in enterprise AI systems. How would you design an incident response plan specifically tailored for AI-specific threats like model poisoning or extraction?
Framework Alignment
What to coverHow do you leverage frameworks like NIST AI RMF and MITRE ATLAS to guide your security architecture? Provide an example of how you would map a specific technical control to these frameworks to satisfy audit requirements.
Model Access and Identity
What to coverDiscuss the challenges of managing identity and access for AI agents and model endpoints in a cloud environment like Azure. How do you ensure the principle of least privilege is maintained when agents require access to multiple data sources?
Behavioural practice
Translating Technical Risk to Business Stakeholders
What to coverDescribe a situation where you had to explain a complex AI security risk to non-technical stakeholders. How did you translate the technical threat into business impact to secure buy-in for a necessary security control?
Collaborating Across Silos
What to coverGive an example of how you partnered with platform or data engineering teams to embed security into an AI project lifecycle. What challenges did you face in balancing security requirements with development speed, and how did you resolve them?
Managing Security in Regulated Environments
What to coverTell me about a time you had to implement security controls in a highly regulated environment. How did you ensure compliance while maintaining the agility needed for AI innovation?
Driving Security Maturity
What to coverDescribe a project where you improved the security maturity of an existing AI platform. What was your strategy for identifying gaps, and how did you measure the success of your interventions?
02
Sharpen your knowledge
Topics to prepare
AI Threat Modeling and Adversarial Testing
Review the MITRE ATLAS framework to map common AI attack vectors like prompt injection, model poisoning, and data exfiltration. Exercise: Create a hypothetical threat model for a RAG pipeline, identifying potential entry points and designing specific security controls for each stage of the data flow.
AI Security Governance and Regulatory Compliance
Study the NIST AI RMF and the EU AI Act to understand how to align technical security controls with organizational governance. Exercise: Draft a checklist for assessing AI model deployment readiness, ensuring it covers data privacy, model robustness, and auditability requirements.
Skills to revise
Secure RAG and LLM Integration Design
Focus on securing the interaction between LLMs and external data sources. Exercise: Design a secure architecture pattern for a RAG system that includes input sanitization, output filtering, and strict IAM policies to prevent unauthorized data access or prompt injection.
AI Security Control Automation
Develop scripts to automate security monitoring and control validation for AI platforms. Exercise: Write a Python script that simulates a basic security check on a model API endpoint, such as verifying that input prompts do not contain prohibited keywords or patterns.
Enterprise AI Platform Security
Gain familiarity with security features in platforms like Azure AI Services, Databricks, or Snowflake. Exercise: Research and document the native security controls available in these platforms for identity management, logging, and data encryption to prepare for architecture assurance discussions.
04
Your preparation checklist
Develop an AI Threat Modeling Framework
Prepare a structured approach to threat modeling for a RAG pipeline or AI agent. Document how you would identify and mitigate risks like prompt injection, data leakage, and insecure tool use, referencing frameworks such as OWASP LLM Top 10 and MITRE ATLAS.
Prepare an AI Security Control Case Study
Create a brief walkthrough of a security control you designed or implemented for an enterprise AI platform. Focus on how you balanced security requirements with operational needs, specifically mentioning how you handled model access, logging, or monitoring in a cloud environment like Azure or Databricks.
Map Regulatory Requirements to Technical Controls
Review the NIST AI RMF and EU AI Act expectations. Be ready to explain how you translate these high-level regulatory requirements into concrete, actionable technical controls for engineering teams, such as automated guardrails or secure API configurations.
Draft Stakeholder Communication Strategies
Prepare examples of how you have explained complex AI security risks to non-technical stakeholders or senior leadership. Focus on your ability to convert technical vulnerabilities into business-relevant risk metrics and practical mitigation plans.
Formulate Targeted Questions for the Interviewers
Prepare questions regarding the current AI security maturity at Societe Generale, the specific platforms (e.g., DATALABS or AI Foundry) they are prioritizing, and how the security team collaborates with data science and platform engineering teams to operationalize security.
Availability has not yet been verified. Status can change after a check; confirm on the employer’s website. Report an issue · Our checking policy
Share this opportunityKnow someone who would be a great fit?