Senior AI / ML Engineer

What You Will Do

• Own the AI layer of our SOC platform: cloud models (OpenAI, Gemini, Anthropic) and local models (Ollama) behind one interface
• Build evaluation sets from real security alerts and measure AI accuracy, cost and hallucination rates
• Build grounding checks so AI-written incident summaries only state facts that are in the evidence
• Defend the pipeline against prompt injection hidden in log data
• Improve anomaly detection (UEBA) and risk scoring, and ship new AI features for analysts
• Keep the service working when AI is switched off
• Red-team our own AI (prompt injection, data leakage) and build AI support for threat hunting
• Desired Candidate Profile 5+ years in ML / AI engineering, including 2+ years on production LLM applications
• Strong Python engineering: tests, code review, version control, APIs
• Hands-on experience with LLM evaluation and with making model output reliable
• Knowledge of attack techniques, threat hunting, vulnerabilities and defence methods, enough to judge AI verdicts on security alerts
• Good understanding of AI in attacks and red teaming: AI-written phishing, deepfakes, AI-assisted malware and reconnaissance, prompt injection, MITRE ATLAS, LLM red-teaming
• Strong written and verbal English skills
• Nice to have Security or SOC domain experience
• Local model serving, quantisation and GPU sizing
• Property-based testing; RAG; AI-security (OWASP Top 10 for LLMs)