"Why should I care about Math? If I want to drive a car, I don't need to know how the engine works"

Build depth, not fluff

"I know how to run ML models without knowing what happens underneath them. So why should I care about the math behind these models? I don't have to know how a car engine works to drive a car." - internet folks.

This is an interesting idea to think. If running an ML model to solve a problem is easy, why should you undergo the pain of knowing the foundations behind these models?

But think about this in another way. If running an ML model is so easy for you it is also easy for almost everyone. So where is the value that you can add to an organization?

Because hundreds of millions of other people also know how to drive a car. But only very few people can drive an F1 car. And yes, F1 drivers do know how their car works. Complex ML models are like F1 cars.

If you want to be a cab driver equivalent in the ML world, just knowing how to run some models might be enough. But if you ever want to do something that not many others can, you need to know the foundations behind the ML models.

Foundational knowledge transcends time. It makes you so confident no matter what the noise around you sounds like.

Irrespective of how much AI changes, these fundamentals remain the same as shown:

Is it enough that you know these topics alone?

So does it mean that if you know just these topics alone, you know AI? Absolutely not. But without knowing these topics in depth, you will never gain the confidence to be a serious ML engineer.

Anything good in life happens at the end of hard work.

Anything sustainable and long term takes time to build.

If you want to become a serious ML engineer, build strong foundations first rather than focusing on toy Kaggle projects.

You need to know how these concepts are used in building AI models.

If you wish to learn the foundations for ML in depth, I am (Dr. Sreedath Panat, MIT PhD) teaching "Foundations for Machine Learning" live, online, starting in 2 days (Jan 25th, 2025) where I will be covering these topics in detail. This is a certified, 4-month course.

The course will start from the very basics, assuming very little prerequisites. The only demand is your time commitment.

If you want to join the live cohort which will be closing soon, you can register here:

One more question: “Why exactly are these topics relevant for ML?”

Here is why.

Here are the topics we will cover as part of the course: