Adversarial Machine Learning Engineer presso C-Serv

Description

The Opportunity

We are building a dedicated AI Red Team to rigorously test and harden enterprise-scale AI products.

We are looking for an adversarial machine learning specialist who thinks like an attacker.

This role focuses on identifying vulnerabilities in LLM-driven systems, breaking model guardrails, exploiting data pathways, and stress-testing AI deployments before they reach enterprise customers.

This is a hands-on technical role at the core of AI security.

What You’ll Do

• Conduct adversarial testing across LLM and AI-based systems
• Execute real-world attack simulations, including:
• Prompt injection
• Jailbreaking and guardrail bypass
• Data exfiltration attempts
• Model inversion and evasion techniques
• RAG manipulation
• Develop scripts and tooling to automate attack scenarios
• Analyse model behaviour under adversarial pressure
• Identify systemic vulnerabilities in:
• APIs
• Embedding pipelines
• Vector databases
• Fine-tuned model implementations
• Collaborate with engineering teams to validate remediation
• Document findings clearly and concisely

You will help ensure AI systems are resilient before they are deployed at scale.

Requirements

What We’re Looking For

Core Technical Skills

• Strong experience in adversarial ML or AI security research
• Experience working with LLM-based systems (OpenAI, Anthropic, open-source models, etc.)
• Deep understanding of:
• Prompt injection techniques
• Model jailbreak methodologies
• AI system exploitation vectors
• Strong Python skills
• Experience building custom attack tooling or experimentation frameworks

AI Systems Knowledge

• Familiarity with:
• RAG architectures
• Vector databases
• Model fine-tuning workflows
• API-based model deployments
• Understanding of model safety mechanisms and guardrails

Nice to Have

• Background in cybersecurity or penetration testing
• Familiarity with OWASP LLM Top 10
• Experience working in enterprise environments

Who You Are

• Curious and relentless
• Comfortable thinking like an attacker
• Creative in finding non-obvious vulnerabilities
• Detail-oriented but fast-moving
• Comfortable operating in ambiguity
• Independent but collaborative

You don’t just run test cases — you design new ones.

Benefits

• Comprehensive Private Medical Coverage
• Support for Mental Health Expenses
• Life Insurance Options
• Attractive Compensation Package