AI Security Researcher II at Microsoft
Redmond, WA
—
Jul 28, 2026
Redmond, WA
Jul 29, 2026
What this job asks for AI summary
A research role focused on the security of agentic AI systems — studying adversarial behavior, failure modes, and security risks in autonomous, tool-using, and multi-agent AI. The position involves designing experiments, evaluations, and benchmarks, then translating findings into practical mitigations and security guidance for real-world AI products. It suits candidates with a research background in AI security, adversarial ML, or LLM evaluation.
Mid level · 2+ years
“or” means any one of them counts — you don't need all of them.
We read this from the posting text with AI. Skim the description below before ruling yourself out.
How this req sits in the market our data
What the occupation pays Median $122,874 (middle half $87,544–$162,374).
Estimated from BLS employment for this occupation and area, per-skill prevalence across our listing corpus, and published wage benchmarks — as of Jul 29, 2026. It is a model, not a headcount.
Why we read it this way (7)
Location is not stated anywhere in the posting — CBSA, state, and remote status are unknown; remote=false is a conservative default only.
The degree requirement accepts equivalent experience in lieu of any formal degree, so the degree requirement is set to None.
The minimum experience threshold is 2 years (with a Bachelor's) or 1 year (with a Master's); the overall years minimum is set to the lower bound of 2 years as stated for the most common path.
Several required skills ('AI security', 'adversarial machine learning', 'agentic AI systems', 'LLMs', 'experiment design', 'benchmarking', 'multi-agent systems') are research competencies rather than discrete named tools; they are included because the posting explicitly gates on demonstrated output in these areas.
No compensation, employment type, or security clearance information is provided.
SOC classification is a genuine toss-up: the role is framed as research (15-2051 Data Scientists) but its primary subject matter is security analysis of AI systems (15-1212 Information Security Analysts); 15-2051 is chosen because the day-to-day work centers on designing experiments, evaluations, and benchmarks — a research/data-science workflow — rather than operational security monitoring or incident response.
Ignored 1 non-technology phrase(s) as skills (responsibilities/concepts, not named tools): experiment design.
Read the full posting
The employer publishes the full description on their own site — read it there ↗. Or sign in to read it here — it's free, and it also lets you track this application.