Senior AI Security Engineer

This company uses AI to change how new medicines are discovered. Their platform runs on autonomous AI agents, right at the edge of what's currently possible, and they need security to keep pace with that speed. Reporting straight into the CISO. Think of yourself as the bridge between ML engineering, platform architecture, and security, making sure nothing sensitive gets exposed as the science moves fast. It’s still a scaling environment. There’s ambiguity, and you’ll be expected to take ownership and shape how security is designed and implemented across the business. The role: As Senior AI Security Engineer, you'll take ownership of security across the AI platform and its agentic systems. Map out the threats specific to AI and ML, and build that into a proper risk framework Decide how model weights, training data, and code get tracked and locked down Build guardrails and sandboxing for LLMs and autonomous agents, with real-time monitoring behind it Work closely with ML researchers and engineers, embedding security across the whole ML lifecycle Lead the response when something goes wrong, using ML techniques to catch what standard tools miss Help turn emerging AI regulation into something the business can actually keep up with Partner with Legal and Compliance to shape what responsible AI security looks like in practice About you (skills/experience): You will bring A genuine understanding of deep learning (JAX, PyTorch, or TensorFlow), and you've worked with large-scale cloud training or inference infrastructure Think like an attacker on AI systems: prompt injection, model inversion, and data poisoning are all familiar, and you know the OWASP Top 10 for LLMs and MITRE ATLAS Proven experience in securing agentic systems before, and understand identity and access controls between agents Comfortable in cloud security, GCP ideally, with solid container and multi-cloud experience Can write production-grade code, Python preferred, so you build your own tooling Communicate well, sit comfortably with ambiguity, and can turn ML risk into something engineers can act on Nice to have: Red-teaming LLMs, agent networks, or ML backends Background in BioTech, Pharma, or Deep Tech Degree in Computer Science, Machine Learning, Cybersecurity, or similar OSCP or a cloud security certification What they can offer: Market-leading compensation (bonus + equity) Hybrid working (3 days in office) Keywords: AI Safety, AI Security, LLM Security, Agentic Security, Prompt Injection, Adversarial ML, MITRE ATLAS, GCP, Cloud Security, Python, Model Risk, Machine Learning, Cybersecurity #J-18808-Ljbffr