AI/ML Security Engineer

Cybersecurity Mid-level (L4 equivalent) 2-5 years

1 Which company are you aiming at?

To generate precise rejection-prevention feedback and application strategy, tell us where you're targeting.

Google

Microsoft

CrowdStrike

Palo Alto Networks

Cisco

Okta

2 Send your application

You need an account so we know where to send the feedback.

Create an account Sign in

The job description

Tech stack. LLM red teaming tools, prompt injection test suites, model supply chain scanning, adversarial ML techniques, Python, PyTorch or TensorFlow basics, data pipeline security, evaluation harnesses

About the role

You will secure the artificial intelligence systems of a technology company shipping LLM-powered products to production customers at scale. The team treats models as first-class attack surface: prompt injection, training data poisoning, model theft, and insecure agentic workflows that take real-world actions. You will red-team features, build defensive evaluations, and define secure patterns for AI architecture. This role matters because AI features are now the front door of the product, traditional application security tooling does not catch model-specific attacks, and your testing is what stands between a cleverly crafted prompt and a serious security incident. You will also build the evaluation harness that regression-tests every model release for prompt injection and data exfiltration, making AI security measurable instead of anecdotal.

What you will achieve

Red-team every LLM-powered feature before launch, documenting jailbreak and prompt injection findings with reproducible proofs of concept developers can verify.
Build guardrail evaluations blocking 95 percent of known injection patterns while preserving legitimate user experience, measured against held-out test sets.
Secure the model supply chain by scanning training data sources and third-party model artifacts for tampering, backdoors, or unexpected behavior.
Define the secure agent architecture covering tool permissions, execution sandboxing, and output validation for autonomous multi-step workflows.
Ship an AI security testing toolkit that product teams run in CI before each model, prompt, or retrieval change.

What you will bring

Must-haves

2 to 5 years in application security, ML engineering, or a hybrid role with genuine hands-on LLM experience.
Practical understanding of prompt injection, jailbreaking, and indirect prompt attack classes with concrete real-world examples.
Familiarity with LLM application stacks: retrieval-augmented pipelines, tool calling, and agent frameworks.
Ability to write Python to build tests, harnesses, and repeatable evaluation pipelines.
Knowledge of data security fundamentals: access controls, PII handling, and data lineage within ML systems.
Clear communication of novel, fast-moving risks to product teams shipping at AI speed.
Intellectual honesty about what guardrails can and cannot guarantee against determined adversaries.

Nice-to-haves

Experience with adversarial ML: evasion attacks, data poisoning, or model extraction.
Familiarity with the OWASP Top 10 for LLM Applications.
Published AI security research or detailed public red team writeups.
Background in traditional application security for defense-in-depth perspective.