Security Architect for AI Research Infrastructure

MATS Research; Greater London; Hybrid; GBP 110,000 - 160,000; Full time

Benefits offered by this job

Lunch provided

Professional development

Flexible PTO for Berkeley-based

Hybrid work flexibility

Job summary

MATS Research is seeking a Security Architect to secure our research pipeline across compute and cloud environments. You will define secure CI/CD, automate compliance, and enable agents to operate safely at scale.

You will own multi-cloud security, cluster hardening, and data protection for sensitive model weights in a fast-paced research setting. You will collaborate with compute and engineering teams to build a secure by-default environment that supports rapid experimentation while preventing

Qualifications

• Deep Linux security experience on HPC infrastructure.
• Terraform and CI/CD with GitHub Actions and GitLab CI in IaC.
• Cloud security for AWS, GCP, and neocloud platforms.
• Hands-on HPC scheduler security (SLURM).
• Threat modelling against adversaries and robust systems.
• Explain security trade-offs to researchers and leadership.
• Sandboxing and least-privilege execution for automated workloads.

Responsibilities

• Harden DevSecOps and CI/CD with security guardrails and compliance as code.
• Secure the software supply chain with SBOMs and vulnerability scanning.
• Define agent sandboxing and least-privilege models for research agents.
• Own multi-cloud and cluster security across providers.
• Harden the research cluster (Linux OS, kernel patching, PAM).
• Protect data and model weights with robust access controls.

Skills

Linux security

Terraform

CI/CD

GitHub Actions

GitLab CI

CI/CD tooling

Tools

AWS

GCP

Neocloud

Job description

MATS Research is seeking a Security Architect to secure our research pipeline across compute and cloud environments. You will define secure CI/CD, automate compliance, and enable agents to operate safely at scale.

You will own multi-cloud security, cluster hardening, and data protection for sensitive model weights in a fast-paced research setting. You will collaborate with compute and engineering teams to build a secure by-default environment that supports rapid experimentation while preventing