Caledonian AI Safety is hiring a Senior AI Trust & Safety Engineer — red team & policy on the Trust & safety team (Hybrid · Edinburgh, United Kingdom). This is a hands-on role in LLM & generative AI—not a generic “AI enthusiast” post.
You will own LLM features, retrieval, agents, and evaluation end to end: problem framing, offline evaluation, safe rollout, and iteration from production signals. Success is measured with clear metrics (quality, latency, cost, or business KPIs)—not slide decks alone.
Caledonian AI Safety supports generative assistants used by tens of thousands of professionals. We obsess over misuse modes, resilient policies, and fast incident containment.
Join us in Edinburgh / hybrid UK as an AI Trust & Safety Engineer bridging offensive testing, tooling, and policy implementation.
What you will do
- Maintain taxonomy-driven red teaming for prompts, plugins, retrieval sources, and file uploads.
- Automate scanning in CI/CD for regressions across policy tiers and locale variants.
- Partner with Legal and Privacy on DPIA artefacts, DPIA refresh triggers, and user-facing disclaimers.
- Run tabletop exercises with support and infra; improve detection and alerting for anomalies.
- Train PMs and designers on guardrail patterns that balance UX with safety outcomes.
What we look for
- 3+ years combining security/offensive mindset with software delivery (Python preferred).
- Hands-on experience testing LLM or chatbot surfaces; familiarity with OSS jailbreak benchmarks.
- Structured communication under pressure; ability to produce exec-ready summaries.
- Comfort reading technical model cards and interpreting limitation statements.
- UK working rights.
Nice to have
- Certs such as OSCP/OSWE or comparable app-sec discipline.
- Experience with SOC2/GDPR-adjacent control environments.