A somewhat disappointing and misleading followup to the excellent first course in this specialization. It's heavily focused on shallow learning on structured data, which is not at all what I think of when I think of the challenges in prod ML.

TFX feels more like a solution to technologies that were available well before the deep learning revolution. There are certainly some useful, albeit complicated, tools coming out of google/tensorflow. We'll see if TFX sticks or just becomes another tensorflow.estimator in the shadow of keras.

I really enjoyed the first course in the specialization but the instruction for the 2nd course (ML Data Life Cycle) was terrible. There were too many concepts packed and explained in an abstract fashion.

I am Technical PM at Amazon and work on ML systems. I was hoping to learn best practices with practical examples. I got a lot of it in the first course but this one was a mash of abstract academic concepts with very few practical examples.

It was irritating to go through the TFX sales pitch without explaining why exactly we are doing something in the ML pipeline.

I did learn some new concepts around auto-labelling and feature selection but I can teach the useful portion of this course in 60 minutes.

I urge Deep Learning AI to replace and rethink this course entirely.

Too many concepts packed in the lecture videos. The lecturor basically reads the slide instead of going through them.

I have been using TF in a professional job for three years. I still find TF is too complex to used.

The labs and assignments were interesting but the lectures, content videos were not engaging.

Most of the course feels like an advertisement for Tensorflow Extended data pipeline management tool. If you are using TF then the tool may be a necessity in some cases, but otherwise it is largely not useful. There is very little which is transferrable outside TF. For the parts which are high level and not TF specific either you know it because you've encountered production systems before, or you don't know it, in which case this course is not really going to help you that much to ramp up.

This course is not up to the usual level of Andrew Ng's specializations on Coursera. In my opinion, it needs a big review to order contents better. The order in which topics are covered sometimes feels like it's just random. You see a subject in week one, video 4 and you see it again at the end of week 2, presented differently. It feels very weird, and if you're trying to summarise the class in a document, it's very hard. Also, there is surprisingly often no link from one slide to the next. It's just "let's change subject and speak about this now". Some quiz questions (for instance in week 2) are referring to material learned AFTER the quiz itself. Also, N-th level negation questions in quizzes are just a way of having us loose our time, not a clever way of checking if some material is well understood. The overall impression this course gave me is that if I had no knowledge of any of this before, it would unfortuntaley not have helped much. The course in general demotivated me into following it, because I was expecting something more interesting and better structured.

I have no doubt in Robert's knowledge on the subject, but delivering clear instruction with just right amount of contexts is an art that takes another few years to master. Way to go!

Access to the code is not available. Most of the concepts are too complicated in implementation. Having used model management before, i think many things should be made much simpler and developer friendlier.

Sadly the lectures are rather dull. But take heart, the material is much more interesting in the next course. Press on. Get this one over with. You will be happier in course #3.

I'm quite dissapointed by this continuation of the otherwise excellent Andrew Ng specialization.

I was expecting a course on frameworks and best practices for managing data in MLOps environments. Instead, this course is basically a commercial of Tensor Flow Extended, a MLOps framework by Google. Other tools often used in commercial applications (like cloud ML platforms) are not even mentioned.

It's true that the course does provide some tips, but they are often too general to be of practical use, specially for people with some experience in the field (e.g. "you need to validate your inputs").

I hope the next courses in the specialization are better.

If you're new to ML pipelines this is an excellent course to understand ML pipelines. Moreover, the labs and assignment are of good quality. If you already are familiar with ML pipelines tech like Amazon sagemaker then this course might seem repetitive of many things but still you get to learn about google's Ml pipeline stack which is TFX.

This handful course allows me to understand how the tft works and how to inspect with statistic aspect of view about data. Much interesting is the practice, it offers much practical example about data preparation, especially the optional week 4 time series data !

This was quite exciting! A lot of new, innovative content in the TFX libraries along with all the theoretical background necessary for determining when to use each component in the data life-cycle, highly recommend!

Unfortunately, the Tensorflow documentation and guides would be much better here.

I felt like the teacher liked to show his ego, it was cringy at some times, and superficial at others. I felt like the entire course could be just put into one notebook, and read in 2 hours. It is a shame as I expected better.

A lot of the concepts are not well-explained. I feel like my mind is constantly drifting away when watching the video, to me, this course is more like a workshop/ads for tensorflow rather than explaining the data lifecycle properly.

I am kind of disappointed with this class. First, I feel this class is way too oriented towards structured data. But more importantly, I find that the labs are overly simple and I can't say that I would feel proficient applying the concepts learned in production.

Tensorflow is not a convinient tool

The course is poorly prepared and presented. The instructor basically talks through slides with no concrete technical content, simply babling from one bullet point to another, from one slide to the next, unorganized. Lectures were horrible -- broad, technical content barely scratches the surface, uninteresting way to deliver and speak. This is a practical course. The intructor should have structured the lecture around a practical implementation through a real-life example. It's not there at all. Very difficult to continue listening and it's very frustrating. Lab and Assignments in Jupyter Notebook are good. Overall, a huge disappointment considering the first course in the Specialization taught by Andrew Ng was so good.

The course takes an overall look over the general data life-cycle pipeline in production. That includes: (1) data collection, labeling and validation; (2) feature engineering, transformation and selection; and (3) data journey and storage. The instructor, Robert Crowe (TF engineer), presented a plain domain of the studied subjects and was fully able to explain them understandably. The technologies and libraries presented through the course are modern and applicable to the majority of my current projects. I would recommend this course to anyone interested into better understanding the data behavior in the production environment, as well as, how to use the introduce libraries to correct data anomalies/problems (e.g. data skew, data drift, others).

Great course. Looking for one on TFX since the tool was open sourced few years ago. While TFX could be quite technical and hard to undertsand, Robert may it clear with many examples to practice and better understand it. Data Scientist that plan to deploy model in production should take it.

This is to understand that Data Lifecycle is the rest of the iceberg, compared to Machine Learning Models being the tip of the iceberg. It is very good demonstration of TensorFlow capabilities processing and maintaining the data for operation.

This is the best course for understanding the data lifecycle in production, everything has been explained in video and the assignments given in the course are the real life practical scenario for data pipeline management for machine learning.

Interesting material. There are quite a lot of typos and many code snippets are directly from the tfx manual pages however the instructions provided and logic of the course is clear.

Thank you so much to DeepLearning.AI. You inspire me! This course is a key step that most part of enterprises should follow in order to construct robust ML systems

Incredibly useful and well teached. Awesome hands-on guide delivered by Robert Crowe, he is indeed a master. Im looking foward to learn with him again!