CrowdStrike
AI Research Scientist, Agentic Systems (Remote)
$120K–$180K
USA - Remote · Full-time · Posted Aug 6, 2026
- Education
- A doctorate is not required; equivalent PhD-level mastery is expected.
About the job
Researches, trains, and evaluates LLMs and agentic systems to automate cybersecurity analyst workflows and improve performance on real security tasks.
What you'll do
- Post-train LLMs and agents, and develop agent workflows for cybersecurity procedures.
- Research agentic planning methods and build rigorous evaluations and benchmarks.
- Collaborate across teams to move prototypes toward production and identify research priorities.
What they ask for
- Strong foundations in machine learning, probability, statistics, and experimental design, with PhD-level research mastery or equivalent.
- Experience training generative models and strong knowledge of LLM training fundamentals.
- Core experience with reinforcement learning/post-training and building agentic systems, including evaluations and prompt optimization.
- Fluency with GPUs, PyTorch, LLM training and serving tools, and reproducible Python research engineering.
Nice to have
- Experience with synthetic training data, agent environments, inference-time scaling, agent safety, or interpretability.
- Open-source contributions, strong technical writing, or cybersecurity knowledge or interest.
Pay
$120,000 - $180,000 per year USD (base salary)
Tools
PyTorch, Hugging Face Transformers/TRL/PEFT, DeepSpeed/FSDP, vLLM/TGI/SGLang, Python, PPO/GRPO/DPO, RLHF/RLAIF
AI focus
Post-training LLMs and agents, including reinforcement learning, agent planning and tool use, and rigorous evaluation for cybersecurity tasks.
Model-written summary · the original posting is authoritative