AI Security Tester

Mastek; Reading; Hybrid; GBP 70,000 - 110,000; Full time

Job summary

Mastek is seeking an AI Security Tester to break and defend AI systems, combining offensive security and AI/ML risk awareness. You will simulate attacks, test guardrails, and identify data privacy risks to secure deployments.

The role requires 5+ years in offensive cybersecurity with at least 2 years in AI/ML security, strong Python/Java skills, and hands-on experience with PyTorch and TensorFlow. Hybrid work with occasional travel to offices or client sites is expected.

Qualifications

• 5+ years in offensive cybersecurity (AppSec, Pen Testing, Red Teaming).
• Minimum 2 years focused on AI/ML security.
• Strong programming skills in Python and Java.
• Hands-on experience with ML frameworks PyTorch and TensorFlow.
• Familiarity with AI tooling LangChain, LlamaIndex, red-teaming toolkits.
• Proven ability to assess and secure AI systems end-to-end.

Responsibilities

• Conduct adversarial testing to challenge AI model safety and guardrails.
• Perform penetration testing and fuzzing of AI systems.
• Identify risks such as data poisoning, model evasion, hallucinations, and bias.
• Validate data privacy controls, including PII handling and encryption.
• Simulate attack scenarios and exploit attempts
• Document vulnerabilities, unexpected behaviours, and evidence in lower environments
• Analyse threat vectors and recommend mitigation strategies
• Proven ability to assess and secure AI systems end-to-end

Skills

Offensive security

AI/ML security

Python

Java

PyTorch

TensorFlow

LangChain

Red Teaming

Data privacy

Education

Master’s degree in CS or Cybersecurity

CISSP certification

Tools

Red-teaming tools

Fuzzing tools

Threat modelling tools

Job description

Are you passionate about breaking AI systems before adversaries do? We’re looking for an AI Security Tester with deep offensive security expertise to help safeguard next-generation AI solutions.

In this role, you’ll simulate real-world attacks to uncover vulnerabilities in AI/ML systems. From stress-testing model guardrails to identifying data privacy risks, you’ll play a critical role in ensuring secure and trustworthy AI deployments.

Working Model

Hybrid Role, when needed will have to travel to Mastek offices or customer location

• Conduct adversarial testing to challenge AI model safety and guardrails
• Perform penetration testing and fuzzing of AI systems
• Identify risks such as data poisoning, model evasion, hallucinations, and bias
• Validate data privacy controls, including PII handling and encryption
• Simulate attack scenarios and exploit attempts
• Document vulnerabilities, unexpected behaviours, and evidence in lower environments
• Analyse threat vectors and recommend mitigation strategies
• 5+ years in offensive cybersecurity (AppSec, Pen Testing, Red Teaming)
• Minimum 2 years focused on AI/ML security
• Strong programming skills in Python / Java
• Hands-on experience with ML frameworks (e.g., PyTorch, TensorFlow)
• Familiarity with AI tooling (e.g., LangChain, LlamaIndex, red-teaming toolkits)
• Proven ability to assess and secure AI systems end-to-end

Nice to Have

• Master’s degree in Computer Science, Cybersecurity, AI/ML, or related field
• Certifications such as CISSP or AI security/safety credentials