Help needed.
Welcome to the forum!
Do you have a specific question?
Thank you.
What is the best and easiest route to getting a remote position as a machine learning engineer after completing this course?
There isn’t really a single defined or “best” route to getting a remote position as a machine learning engineer in for any other type of position in general. Different people reach those roles through different paths depending on their background, experience, and the opportunities they find along the way.
The most effective approach is to focus on building your skills incrementally. Start with the fundamentals if you are not already comfortable with them—especially Python, and the core mathematics used in AI such as linear algebra, probability, and statistics.
On this platform, there are courses covering:
- the fundamentals of machine learning and AI
- more advanced deep learning topics
- and many application-focused courses across different domains
It’s usually better to combine learning with practice. Try working with real datasets from platforms like Kaggle or other open data sources, and build your own small AI projects. This helps you move from theory to practical understanding.
Over time, as you build consistency in learning and project experience, you’ll be in a much stronger position to apply for roles, including remote machine learning engineer positions.
I don’t understand why Andrew says the NST cost function can be computed in a shallow layer, a deep layer or anywhere in between.
The cost function is computed ONLY at the final output of the network surely.
Neural Style Transfer is a very different case than the ML models we’ve seen up to this point, which are typically doing classification (making a prediction based on the input data). The goal here is completely different in that the model is trying to sort of “blend” two different images and use the artistic style of one to modify the other. So we need a cost function that expresses that goal in some way. The goal here is aesthetic, which means it is fundamentally subjective, as opposed to the classification case in which there is a defined right answer. So this situation is totally different. Since there is no correct answer here, we can try experiments and see how the cost function we define affects the results until we find a function that gives results that are visually pleasing. How can we capture or express information about the artistic style of an input image? How would you even define “style” if you had to express it as a mathematical function? As opposed to “is this a picture of a cat or not?” 
With those thoughts in mind, it might be worth listening to the lectures again and see if what Professor Ng says “lands” in a different way for you.
BTW that’s a completely different topic than your previous question, so it might make sense to have created a different thread with a more specific title. The contents of the forum can have value for future students who have the same or similar questions, so it’s worth making things easier to find.
This question is off-topic for the thread (about employment), and a new topic should be created for it. Community members and staff who are familiar with ML and AI employment practices will be able to find it more easily.
I have moved the thread to the “AI Discussion - Careers” topic, since employment opportunities are more appropriate for that area.