Enjoyed the course and it is very interesting. Although there is no formal "prerequisite" for the course, you will get much more if you have various basic concepts in AI/ML, python, Jupyter notebook, CI/CD & Google Cloud Build, K8S & GKE, YAML, Github - especially for the labs, I really enjoy them. You might see some people saying that they hit minor problems - in fact, those minor problems are also part of the learning.

It was amazing course and content. No doubt that you are best content provider for the study material. you are feeling the gap between industry and university. As a learner i also faced some difficulty which you need to review it once in "QUICKLABS" cluster creation.

THANKS :)

Ronit Sagar

I learnt new concepts in machine learning through google cloud platform and i am so happy for that. Thank you Coursera for giving this opportunity to gain Google certification and i learnt a lot about google cloud, Kubeflow, and had practical experience through graded external tool.

Good course for overall view of Kubeflow orchestration, basics of kubeflow and containerisations and ML ops services available on GCP. Highly recommended if you wanna deploy models with best practices!

Loved the content, labs, and regularly intervened quiz. The only suggestion is that, for Juniper Labs, a detailed video solution would have added more value to this course.

I think there should be more content about AIML can be better choice or preferable.

Otherwise all the things are okay I enjoyed this course and learn a lot.

ThankYou So much.

It is a good designed course, but I would prefer to have basic knowledge of Machine learning and data science in order to understand this course even much better.

A few of the labs didn't work, had to contact support. Also would be good to point to documentation for various tools like kfp-cli

Otherwise good.

Content well written. Some lab issues. Resolved but frustrating. Language processing a bit off on transcribed material from speakers.

The course is quite educational, yet the lab material can sometimes be confusing, especially for beginner users

I​t's ok. There are example notebooks to understand the code. The pricing part is missing.

It was good experience learning about the deployment process of ML application on GCP.

I miss a chapter on automating monitoring models when metrics diverge

Needs to cover the subject in greater detail

It's a good tutorial for beginner

Good learning experience.

The labs are so important

learned something new

Material is good

Good Course

while this course teaches some useful skills, in particular how to to offload ML workloads to GCP, and introduces Kubeflow well, it doesn't go into enough depth to really let the students master the material. It doesn't help that Kubeflow (and its GCP implementation) are fundamentally fairly complicated technologies that compete with other, more mature (but less specialized) tools like Airflow. All in all, a good starting point, but don't expect to master the material - further study will be required. This course only scratches the surface.

The course gives a nice overview, but either it should be more generic and fun, or more detailed and techy but also longer. Now it feels like its trying to do both and failing at it. It is a bit too condensed and boring on the practical parts, and most of the tasks can be solved with copy paste, and somehow I don't feel that the whole course motivated me into stop copy-pasting and instead actually learn these things. Several of the Qliklab workshops seem to be buggy.

The content is well designed and explained. The Hands-on Lab sessions need a lot of improvement. MLOps is implemented in a really complex manner (but that is more about a comparison between GCP and other providers). But for ramping up MLOps on GCP, this course is a really good starting point. Best of Luck!

I took this so I could get better at Kubeflow on EKS (not Google Cloud) and it was not worth it. The Beginning is promising and the explanation of kubernetes was great, but then it quickly became not applicable. If you are using it for GCP then it may be worth while.

The labs are too basic, I expected to view how to use tools such as tensorboard with kfp, with the intention to track progress of the models. But more relevant is the lack of examples on how to train/hyperparameter-tunning using a kfp alone avoiding AI jobs tool.