AI Security Architect – Cybersecurity, Île-de-France

Stack

Responsibilities

• ▹Design secure target architectures for machine learning, generative AI (LLM, RAG, copilots) and agentic AI (multi-agent, autonomous workflows) systems
• ▹Define and deploy MLOps/LLMOps pipelines: continuous integration, model monitoring, lifecycle management (versioning, retraining, decommissioning) and AI model security tooling
• ▹Take part in solution choices (AI cloud, LLM platforms, orchestration frameworks, AISPM, guardrails, data and model leakage protection) and validate directions through POC/POV
• ▹Run architecture reviews of AI systems and present recommendations to technical committees and senior stakeholders
• ▹Qualify AI-specific security needs and define requirements (availability, integrity, confidentiality, traceability)
• ▹Identify and map threats targeting LLM and agentic AI systems (prompt injection, data poisoning, model exfiltration, evasion, jailbreak) using MITRE ATLAS and MITRE ATT&CK
• ▹Integrate security and privacy by design principles into specifications, backlogs and MLOps pipelines
• ▹Manage AI-related vulnerabilities and risks (AI red teaming, adversarial penetration tests) and support remediation with SOC/CERT teams
• ▹Write AI threat analysis reports for both technical analysts and decision-makers
• ▹Conduct risk analyses on AI use cases (EBIOS RM, NIST AI RMF, ISO/IEC 42001 and 27005), maintain the AI risk map and follow treatment plans with GRC teams
• ▹Act as AI and security reference for project teams, promote good practices and help the team build skills
• ▹Contribute to pre-sales activities and proposal building

Requirements

• ▹Degree from an IT engineering school or equivalent
• ▹At least 10 years of cybersecurity experience, notably in architecture-related topics
• ▹Mastery of modern AI architectures: traditional ML, LLM/RAG, agentic systems, cloud platforms (AWS, Azure, GCP) and MLOps tools (MLflow, Kubeflow, etc.)
• ▹Knowledge of solutions usable in a sovereign AI system (Mistral, OVH, etc.)
• ▹In-depth knowledge of AI security frameworks (MITRE ATLAS, OWASP Top 10 for LLM Applications, NIST AI RMF) and risk analysis methodologies (EBIOS RM, ISO 27005/42001)
• ▹Threat modeling applied to AI, red teaming and adversarial penetration testing, and threat intelligence tools (MISP, OpenCTI)
• ▹Excellent communication skills to explain technical topics to C-level decision-makers

Soft skills

What we offer

• ▹Remote-work agreement: up to 2 days a week depending on assignments
• ▹Benefits package: supplementary health insurance, works council (CSE), meal vouchers, profit-sharing agreement, holiday bonuses and referral bonuses
• ▹Multiple career opportunities: more than 30 job families
• ▹Hundreds of trainings to develop skills and progress within the Group
• ▹Self-paced training platforms to prepare for certifications and support personal development
• ▹Possibility to get involved with the company foundation or the partner Vendredi
• ▹Opportunity to join the Tech'Me UP community (training, conferences, tech watch)

About the company

Sopra Steria is a major European tech company offering consulting, services and digital solutions that help clients with their digital transformation. It has 51,000 employees in nearly 30 countries and reported revenue of €5.6 billion in 2025. Its subsidiary Sopra Steria I2S, dedicated to infrastructure, cloud and cybersecurity, has 3,000 employees.