What's the difference between a ML researcher and ML engineer?
Asked on 2020-11-24 03:02 by Lex S.
Both job titles have similar requirements in my opinion. What's the difference?
Response #1By Wojciech G. on 2020-11-24 03:11
The best way to think of this is as the difference between research and application.
ML researchers focus on the model development and experimentation. They do research, read papers, explore new algorithms, all with the ultimate goal of making a model more effective or accurate. Their research focuses on the model, and they worry less about things like resource load, product features, and other "applied" components of their work.
ML engineers focus on productionizing and maintaining the ML system. This includes architecting the architecting the entire solution end-to-end, data processing pipelines, managing server resources, and monitoring/testing everything to ensure it stays performant and hasn't crashed.
The two roles often require very different mindsets. ML researchers tend to be more academic in nature and tend to focus on prototyping, while I tend to prefer that ML engineers be perfectionists when it comes to optimizing and running code/processes.
Most companies make the mistake of hiring many ML researchers when in reality they simply need an ML engineering team. When you have a team of two or three "researchers" working on building models and putting them into a product, then they are often doing the complete engineering work, and sacrifice model performance metrics in doing so.
Note that every company might be slightly different in how it thinks about this. If you are actively interviewing, make sure to ask your hiring manager about the difference, too!