Security Engineer · MITRE ATLAS & Adversarial Defence

Role overview Senior security engineering role focused on protecting AI platforms, agents, and workflows from adversarial threats. The work combines offensive testing with hands-on defensive engineering: designing runtime guardrails, addressing prompt injection, protecting models, and translating MITRE ATLAS techniques into enforceable controls. The role is listed as London-based with remote flexibility and is part of a trust and compliance function.

Responsibilities - Design layered protections spanning pattern detection, sensitive-data redaction, tenant-level policy, and model-based safety controls. - Convert MITRE ATLAS adversarial techniques into practical runtime safeguards rather than relying only on documentation or assurances. - Conduct red-team exercises against agents and workflows, identify weaknesses, and implement fixes. - Collaborate with compliance-engineering colleagues on controls governing safe and accountable AI usage. - Help strengthen defences against prompt injection, model misuse, and other threats affecting agentic systems.

Requirements - Professional experience in security engineering with both offensive and defensive capabilities. - Familiarity with threat models for large language models, AI agents, and related attack paths. - Working knowledge of the MITRE ATLAS framework and its relevance to AI security. - Strong implementation skills and a track record of shipping security controls, not only producing assessments. - An analytical, adversarial mindset and the judgment to distinguish theoretical risks from exploitable weaknesses.

Benefits and work setup - Senior full-time position with London or remote working options. - Remote availability across the UK, EU, and selected global time zones. - Seed-stage equity allocation and cash compensation benchmarked against leading London and remote markets.