Principal Model Optimization Engineer
Core
Optimizing machine learning models for performance on GPU architectures across training and inference workflows for Roblox's ML Platform.
Role type
Principal Model Optimization Engineer
Builds
High-performance ML inference and training pipelines for Roblox's Discovery, Safety, and Engine systems.
Domain
Internet / Machine Learning / GPU Systems
Deliverable
production ML models
Required skills
GPU debugging, CUDA, Triton, TensorRT, LLM optimization (speculative decoding, continuous batching, quantization), system design, low-level performance profiling
Preferred skills
Tooling development, cross-functional collaboration, performance advocacy
Technologies
CUDA, Triton, TensorRT
Responsibilities
Optimize ML models for GPU performance, conduct low-level profiling to identify bottlenecks, develop best practices and tooling for model optimization, integrate optimized models into production, build interfaces for ML platform usability
Seniority
Principal, hands-on IC
