> Markdown version of [/jobs/ext/3739959-ai-security-engineer](https://www.wearedevelopers.com/jobs/ext/3739959-ai-security-engineer). Every page supports `.md` or `Accept: text/markdown`. Links point to the HTML versions so they work for humans too. Agent guide: [/agents.md](https://www.wearedevelopers.com/agents.md).

---

# AI Security Engineer

- **Company:** Bjak
- **Location:** London, UK (Remote available)
- **Experience:** Experienced
- **Salary:** £49,349.0
- **Contract:** Permanent contract
- **Skills:** Application Programming Interfaces (APIs), Artificial Intelligence, Amazon Web Services, Cyber Security, Computer Programming, Customer Data Management, Information Leak Prevention, Python (Programming Language), Node.Js, Role-Based Access Control, AI-enabled Products, TypeScript, Data Logging, Scripting, Retrieval-Augmented Generation, Large Language Models, Software Security, Information Technology, Virtual Agents
- **Published:** October 15, 2026
- **Apply:** https://www.adzuna.co.uk/jobs/details/5925559805

## About the Role

* Degree in Computer Science, Cybersecurity, AI, or related field, or equivalent experience.
 * 3+ years in application security, product security, security engineering, or AI system security.
 * Experience implementing or owning SC TRM and BNM RMiT controls, technology risk, or regulatory control evidence.
 * Understanding of third-party LLM integrations, model APIs, agent tools, RAG, and AI application architectures.
 * Knowledge of prompt injection, indirect injection, cross-user data leakage, excessive agent permissions, and vendor data disclosure risks.
 * Programming or scripting skills, preferably Python and TypeScript/Node.js, plus APIs and AWS or GCP.
 * Able to collaborate with Product, Legal, Compliance, Data, and Engineering on practical controls and risk communication.

English is our main working language across global teams. Strong English communication is required.

## Description

Secure BJAK's AI-enabled products and agent workflows using third-party models. Focus on customer data shared with model providers, agent permissions, user data isolation, and prompt injection through uploaded documents and other untrusted content., * Assess customer data sent to third-party model vendors, including minimization, vendor controls, retention, logging, and approved use.

 * Design controls limiting AI agent permissions, tools, actions, and system access using least privilege and explicit authorization.
 * Implement and test tenant and user data isolation across prompts, conversation history, retrieval, memory, and AI-connected services.
 * Threat model and test prompt injection and indirect injection through uploaded documents, retrieved content, links, and other untrusted inputs.
 * Partner with Product and Engineering on safe document ingestion, content handling, output validation, and approval for sensitive actions.
 * Support SC TRM and BNM RMiT for AI technology risks, including assessments, third-party oversight, control evidence, and remediation.
 * Build security tests, monitoring, and response procedures for AI misuse, data exposure, unauthorized actions, and vendor incidents.