Research Engineer, Post-training Infrastructure

About the Team

Our team is responsible for the “post-training” or alignment of chatGPT. We integrate various improvements from the rest of the company into our RLHF process ultimately producing the models used by millions of users both in the chatGPT product and API.

About the Role

We are looking for an engineer to help improve the efficiency and reliability of our AI model training systems. You will have the opportunity to work across the technology stack, from optimizing low-level components like GPU kernels and network traffic, to developing intuitive front end interfaces.

The ideal candidate will have a strong technical background across areas like data technologies, distributed systems, and writing reliable software. Research experience is not required but experience in some kind of ML environment is (i.e. applied ML roles). This role involves analyzing and troubleshooting complex system issues, implementing solutions, and finding ways to prevent future failures.

This role is based in San Francisco, CA. We use a hybrid work model of 3 days in the office per week and offer relocation assistance to new employees.

In this role, you will:

• Be responsible for unblocking and keeping our systems which power the models behind chatGPT running smoothly.
• Be willing to dive into large ML codebases to debug.
• Be responsible for keeping very large ML training jobs going smoothly.
• Fixing a variety of non-ml software things involving data quality, data prep, job startup speed, CI performance for our team’s tests, and so on.
• Debug and fix issues on the intersection of ML and systems.
• Sample projects include:
• Figuring out why a new cluster suddenly has 10% of experiments fail.
• Diagnosing and fixing regressions in our data pipelines.
• Fixing 30% slowdown in our RLHF training code.

You might thrive in this role if you:

• Are a team player – willing to do a variety of tasks that move the team forward.
• Experience working in complex technical environments
• Experience debugging ML systems.
• Experience with reinforcement learning and or transformers
• Experience with the python
• Experience with kubernetes / distributed infrastructure
• Experience with GPU’s
• Experience with 1 or more large scale data system such as beam or spark.