As a machine learning engineer, do you really need to understand the low-level math behind the models? I came across a fascinating discussion on Reddit that got me thinking about this. A team with no prior machine learning experience was able to achieve good results by simply watching online videos, understanding popular ML models, generating features from raw data, and feeding them into an ML model API. They didn’t need to dive deep into the math behind gradient descent or other complicated formulas.
So, when do you really need to understand the math? In my opinion, it’s when you want to innovate or improve existing models. When you’re working on a project that requires customizing or developing new models, you need to have a solid grasp of the underlying math. Otherwise, you’ll be limited to using pre-built models and APIs without truly understanding how they work.
But what if you’re not working on cutting-edge research or developing new models? Can you still be a successful machine learning engineer without understanding the math? Absolutely. Many companies use pre-built models and APIs, and they need engineers who can implement and fine-tune them effectively. In such cases, understanding the math is nice to have, but not necessarily a requirement.
However, having a basic understanding of the math can still be beneficial, even if you’re not working on research or development. It can help you troubleshoot issues, understand how models work, and make informed decisions about which models to use.
What do you think? Do you need to understand the math to be a successful machine learning engineer? Share your thoughts in the comments below!