An incredibly challenging course with a lot of juicy content. Builds heavily on previous courses in the Data Structures and Algorihms specialization such as hashing, graph searching, data structures, stress testing, algorithmic complexity etc. But given you completed those, you already know how to solve such problems, and it's rewarding to see all the pieces working as you put them together. Also, this course requires some additional maths knowledge such as linear algebra, logic and probability theory. All in all, the "advanced" attribute fits well.
Assignments are wildly varying in difficulty - completing one took me 3 days, another was done in just 30 minutes. They were mostly fine (except for the simplex linear programming solver which I haven't even attempted) and forums were very useful for guidance.
The videos themselves are usually okay, the only seriously lacking area is the linear programming week with Daniel Kane. LP (a fundamental subject in computer science) in itself could fill a whole course, but simply put, his explanations and examples fell short, it was very hard to understand them. So refer to the additional readings there if you are interested. On the other hand, the rest of the course is nice.
Week 1 (flow networks) with Daniel was also kind of hard, but not impossible to understand. Alexander Kulikov's 2 weeks on NP-completeness are the high mark of the course, with engaging and clear explanations. Michael Kapralov's optional 'heavy hitters problem' videos are also interesting and pretty well explained.