Senior ML Engineer, RL & Self-Improving Agents

Fdetimes; Bellevue (WA); On-site; USD 182,000 - 242,000; Full time

Benefits offered by this job

Medical, dental, and vision insurance—

401(k) with generous employer match

Flexible PTO

Tuition Reimbursement

Employee Stock Purchase Program (ESPP)

Paid Parental Leave

Catered lunch daily

Job summary

CoreWeave, The Essential Cloud for AI, is seeking a results-driven ML researcher/engineer to advance self-improving agents. You will generate and evaluate research ideas, validate directions across real customer tasks, and apply substantial GPU compute to push continuous learning in production.

You will work with the CRAFT team to ship impactful solutions and own components from CUDA kernels to model tooling.

Qualifications

• Bachelor’s, Master’s, or PhD in Computer Science, ML, Robotics, or related field
• 4+ years ML experience or PhD with 2+ years in industry, focus on model training
• Strong Python skills; experience with PyTorch or JAX
• Strong understanding of LLM post-training: supervised fine-tuning, RL, on-policy distillation
• Experience deploying ML models in production environments

Responsibilities

• Generate and investigate research ideas toward continuous learning in production
• Validate research directions with real customer tasks across the CRAFT team
• Leverage GPU compute to accelerate experiments
• Touch multiple stack areas from CUDA kernels to LLM tracing dashboards
• Ship practical solutions with fast learning and ownership

Skills

Python

PyTorch

JAX

LLM post-training

Production ML

Education

CS/ML/Robotics (Bachelors/Masters/PhD)

Tools

Distributed training

GPU acceleration

Megatron

CUDA

Job description

CoreWeave, The Essential Cloud for AI, is seeking a results-driven ML researcher/engineer to advance self-improving agents. You will generate and evaluate research ideas, validate directions across real customer tasks, and apply substantial GPU compute to push continuous learning in production.

You will work with the CRAFT team to ship impactful solutions and own components from CUDA kernels to model tooling.