Principal Model Optimization Engineer

Roblox
San Mateo, California, United States
Full Time
Hybrid
Senior
USD 295,250 – 345,040
Bachelor Degree
Technology Information Technology Telecommunications

Posted 1 week ago

Apply to this job and hundreds more automatically

Get Started
Excellent 4.7

Job summary

The role involves optimizing machine learning models for GPU performance, including training and inference, by analyzing and improving pipeline bottlenecks, developing best practices, and collaborating with data scientists and engineers to deploy models efficiently.

Responsibilities

You will set the technical direction for the ML platform by designing scalable GPU and compute management systems to improve training and inference efficiency. You will also lead deep investigations into performance bottlenecks and collaborate with cross-functional teams to enhance platform reliability and developer productivity.

Qualifications

CUDA GPU architectures LLMs low-level performance analysis model optimization performance profiling quantization speculative decoding TensorRT Triton

Candidates must have 6+ years of professional experience with a strong track record in building end-to-end ML infrastructure and managing GPU workloads at scale on Kubernetes. A Bachelor's degree in Computer Science, Computer Engineering, or a related field is required.

English

Benefits

  • Equity compensation
  • Health insurance
  • Reasonable accommodations

Company

Roblox

Roblox

linkedin
Software Development
Location
San Mateo, California, United States
Website
roblox.com

Apply to this job and hundreds more automatically

Get Started
Excellent 4.7
Pete DeQuarto avatar
Pete DeQuarto

AIApply is a total game-changer. Thanks to Auto-Apply, I have been applying to jobs I would never even have come across otherwise!

Akarshak Tanwar avatar
Akarshak Tanwar

In one week I applied to over 150 jobs and locked in 2 confirmed interviews, with a support team that was quick to respond. The auto-apply feature is fascinating and a real pleasure to use!

JC
Julian Carlin

It really worked. In under 48 hours it had applied to 100 jobs on my behalf without any time or effort from me. I am looking forward to seeing how it keeps improving.