Oracle Technical Track

GCP features wish list for Google Next 2018

2 min read

Jul 20, 2018 12:00:00 AM

As I've just posted about Pythian at Google Next, and I though it would be cool to share what are the new GCP features I'd love to see introduced next week. Since I'm focusing on building AI products these days, my wish list is focused there. First of all, I want to see more beta products and features to transition to GA just to be confident using them in production for the clients. Specifically, Google Cloud Functions have been in beta way too long and while I'm on that topic... Google, would you please announce "planned" Python 3 support? Otherwise, it's kinda useless to us... In the area of AI itself... Google Cloud AutoML is at the top of my wish list, especially now that we’ve gotten to work with its alpha release. AutoML has several neural network architectures pre-created for specific classes of ML problems and tunes the network architecture (using ML to train ML as Google says). This means that you can feed in arbitrary pre-labeled objects like images and the service does the rest including converting the objects into model features. Here is how it works and here are more brainy details and here is the demo of AutoML Vision user experience. If Google opens up AutoML at the show next week then hopefully we could show what we did with it analyzing Cricket match recordings for one of our clients. One area where Google has been creating lots of new products is Cloud AI APIs for specific use cases with pre-trained models such as Vision API, Speech-to-Tex API and Video Intelligence API. I suspect we may see new APIs as well as enhancements and new features to the existing APIs. Dialogflow Enterprise Edition seems to be in very high demand these days so conversational UI is a hot area for innovations! I've mentioned Python 3 for Cloud Functions and it would also be good to see its support for Apps Engine Standard. Of course, we can use flexible environments already but orchestrating our custom docker images on Kubernetes Engine seems to be as good or better. Maybe Python 3 will never make it to Apps Engine Standard as Kubernetes is the new direction... Something I would like to see in Stackdriver is evolving it into permanent platform to store logs and metrics instead of the need to export them into the long term / archive storage. The export itself is easy but it required to build some services on top of that for long term access. I'd rather have one user experience and data availability interfaces. One can wish... Finally, while Google Data Studio is pretty cool, it's capabilities are rather simplistic and it's not very polished yet comparing to something like Tableau. I'm really missing a visualization product as a GCP service that's similar in capabilities and ease of use to Tableau. Turning biblical "Ask and it will be given to you". Well, hope at least some! :) Do you have the wish list of GCP features?

On this page

Ready to unlock value from your data?

With Pythian, you can accomplish your data transformation goals and more.

Speak with Pythian consultants now →

What We Do

GCP features wish list for Google Next 2018

Share this

Share this

More resources

How to Deploy Machine Learning on Google Cloud Platform

Log buffer #545: a carnival of the vanities for DBAs

Apache beam pipelines with Scala: part 1 - template

Ready to unlock value from your data?