Subscribing to this course only gives you option to run assignments on Qwik labs, and they're very poor for these kinds of assignments. You won't get any feedback on assignments anyway since there is no grader.

If you want to check out the material it's better to just clone training-data-analyst from github and do these assignments on GCP free tier.

Too short and fast. For people who are not acquainted with cloud platform and other google tools, it would be difficult to understand. Aslo the tutorial videos are also not descriptive.

This is one of the two worst trainings I have taken in the recent 15 years, courses, subjects, units of universities, Coursera, Udemy, Pluralsight, etc.. all included.

Why? A student comes to learn about things. They generally hope, by implication, to be taught by qualified staff who know what they are talking about and who know their courses. These people are not here. The people in the videos are not answering questions. Those who answer questions here don't know much about the course. Students have to give them links, after already telling them section titles. Even when they replied, students would have to try hard to bring them to the course material under discussion - otherwise, they wrote about something else.

Furthermore, they are definitely not subject experts who are qualified answer such highly technical subject matters. They can give students links over the Internet, in most cases. Then, why do students come here? Youtube and Internet would be enough.

Number 3, the course is terribly created, scripts mostly read to students by machine-generated voices. Materials were mostly quite randomly put together. Later materials cannot find earlier discusions for foundational discussions. There were many concepts not explained, discussions where prior foundations could not be found.

If this is not enough, after students have tried to finish the course by the due date, there is a lab that has been denying students' access for over three weeks. Three weeks! Not enough to resolve an issue. I have personally contacted the 'staff' here, Qwiklabs support, Coursera support, multiple times, trying to find a way through or around. None of them cares. They asked students to wait. It's been three weeks!!! The course itself is only three weeks. Students come to this lab at about the end of the second week. The only outcome is that students will have to pay more. I told them of this. How did they help? Nothing. No workaround.

I have gotten so many apologies that I realize they are mostly copied and pasted and sent to me. The small merit this course has is that there are videos and support staff replied, often by scratching the surface.

In short, the course is a perfect representation of the greed, greed, and more greed of the training provider. Productivity, profit and scale of operations are their objectives. Not students' education and welfare.

It is very good course, gives good overview over large ML systems on cloud, a lot of examples from real implementations gives good understunding about problematics in projects realisations

I did not realize the many aspects to consider implementing a Production ML system. This course presents all of them and provides guidance for evaluating alternative

Very theoretical.

I was not able to complete de course because I wasn't able to lauch a lab. I reported the issue but got no answer. Is frustrating that even though I spent a lot of time taking the course, because of the non existing lab, I was not able to receive my certificate. The content is good, but there's no way to go further after the problematic lab

Not just an issue with this course but it happens with many of the Qwiklab labs. Some of these labs demonstrate complex topics but don't go in to the details. You're given a bunch of command line commands to run which are usually about setting up the environment (local & cloud). Then when it comes to the specific sub-topic (why you're setting these variables) it has very little to say. From that point you're left to research the topic yourself and it can take some time to find it in the documentation. It would be far better if there was call out links to the documentation so I don't have to break my flow of learning taking 15-20 mins to find the information I need in the documentation.

Two labs does not work and if you do not contact the support. The course is unfinishable.

A lot of great production examples, labs and reviews but perhaps too many issues for a single course - however I understand that it was perhaps to provide an overview of the possibilities, a kind of "toolbox" for production ML.

Excellent overview of designing real-world ML systems. Some of the labs are daunting, but the emphasis is showing you what can be achieved, rather than achieving mastery within the course.

very good for people who will enter the production stage on machine learning systems

It was bit hard course but lab work was great and learn many production level consideration for ml systems.

Very practical which was nice. Thank you for adding the Quicklabs that helped a lot.

While most of the content is sufficiently informative for a course, the implementation itself has too many issues: wrong videos in some modules, errors in quizzes, and so on. Once they organize the material properly, this course can definitely be 5 stars.

The first module was really good, but the others just seemed like an ad for GCS. Also, the 3rd and 4th module the labs / lab video was hard to follow and felt like I was just reading random code.

Overall this is a very good course.

+ Working as a less experienced data scientist, I gained a lot of hands-on knowledge when putting a machine learning model into production. Especially the dos and don'ts as well as what to pay attention to.

- It would be better to provide more architectural overviews, or further readings, regarding ML on Google Cloud, just for people with less GCP knowledge to catch up easily. Also 2 hour for an exercise is a bit short as I need to hurry up when there is a lot to do in that exercise. Even though I can also start the same exercise 2nd time, it is a lot of hassle to repeat the previous steps. I would suggest that depending on the % of people which can finish the lab, extend the hours for some of the users. After all you have this data and can do some analysis, isn't it? ;)

In general a lot of very valuable knowledge. These are topics that are "hard to teach".

Ways to improve may include the following:

1. more detailed projects and more comments on the code

2. more pictures

3. more correspondence between talk points and slides.

4. A clearer structure that integrates all the topics together.

I really enjoyed the course. It gives the insights on things that are to be taken into consideration while making ML models in production. Though the course focuses on GCP, the learning can be easily applied to other platforms.

This course, I believe, will be crucial to my future understanding of end to end ML applications. I only hope gain more practice in assembling all the pieces from experimentation, to model design, to data engineering.