Microsoft
Member of Technical Staff (Senior AI Security Researcher)
$120K–$261K
United States · On-site · Staff · Full-time · Posted Oct 29, 2026
- Experience
- Bachelor’s: 4+ years; master’s: 3+ years; doctorate: 1+ year; or equivalent experience. Preferred: 2+ years with Python and machine learning frameworks.
- Education
- Bachelor’s, master’s, or doctorate in Statistics, Econometrics, Computer Science, Electrical or Computer Engineering, or related field; or equivalent experience.
- Visa sponsorship
- The position requires passing the Microsoft Cloud Background Check upon hire or transfer and every two years thereafter; other Microsoft, customer and/or government security screening requirements may apply.
About the job
Conducts post-training research on open-weight language models for cybersecurity, including developing recursive self-improvement methods and delivering reusable models and workflows to red-team partners.
What you'll do
- Develop and curate training data, objectives, and experimental controls; assess post-training methods.
- Implement reproducible training pipelines and evaluate model performance, generalization, and security.
- Develop recursive self-improvement methods and deliver trained models and workflows for red-team partners.
What they ask for
- Bachelor’s degree in a stated or related field and 4+ years of related experience, or a master’s degree and 3+ years, or a doctorate and 1+ year, or equivalent experience.
- Ability to meet Microsoft, customer and/or government security screening requirements, including the Microsoft Cloud Background Check.
Nice to have
- Experience training open-weight language models with distributed or multi-GPU infrastructure, or using preference optimization or reinforcement-learning-based post-training.
- Adversarial or security-focused language-model evaluation; relevant publications or released research software; or 2+ years of Python and machine learning frameworks experience.
Pay
USD $119,800 - $234,700 per year (base pay); USD $160,200 - $261,000 per year (base pay)
Tools
open-weight language models, Python, machine learning frameworks, distributed or multi-GPU infrastructure
AI focus
Post-trains language models for cybersecurity tasks, including recursive self-improvement; evaluates training methods and model security.
Model-written summary · the original posting is authoritative