Senior Security Engineer (AI Safety), London, Lausanne

Overview

In this role you will architect and secure an AI-first drug discovery platform at Isomorphic Labs, bridging ML research, platform engineering, and defense. You’ll build risk frameworks for AI/ML in life sciences, protect model artifacts, and embed security into training and inference. You’ll lead AI-specific incident response and drive model-driven defenses to safeguard IP while accelerating innovation. Join a mission to push digital biology forward with rigorous security and collaboration.

Pay / Benefits

• hybrid working

Responsibilities

• Adversarial threat modeling for AI/ML vulnerabilities within an AI-powered life sciences platform
• Establish granular protection for ML model weights, code, and training data
• Design scalable guardrails, sandboxing, and monitoring for the LLM ecosystem (including ADK, MCP)
• Implement security controls across the ML lifecycle from data ingestion to inference
• Lead AI security incident response and automate threat hunting using ML techniques
• Automate risk posture monitoring and compliance metrics aligned with regulatory requirements
• Develop and implement AI safety standards in coordination with Legal, Compliance, and partners

Key requirements

• Deep understanding of deep learning frameworks (e.g., JAX, PyTorch, TensorFlow) and large-scale cloud training/inference
• Familiarity with AI security threat vectors (prompt injection, data poisoning) and frameworks like OWASP Top 10 for LLMs or MITRE ATLAS
• Experience securing agentic frameworks (ADK, MCP) and agent-to-agent identity controls
• Cloud security expertise (GCP preferred), container security, multi-cloud/SaaS integrations, network isolation
• Ability to translate ML risk into actionable engineering tasks for leadership
• Production-grade Python coding for security tooling and policy enforcement
• Strong translational communication to collaborate with AI researchers

• Ambiguity navigation
• Cross-functional collaboration
• Clear technical communication
• JAX
• PyTorch
• TensorFlow

…