François Chollet on Twitter: "Kubernetes is great for deploying ML apps. Here's a detailed guide on how to use Kubernetes to deploy a Keras model as a REST API on Google Cloud: https://t.co/VEWpKTsgGe"

Country	Code	For customers of
United States	40404	(any)
Canada	21212	(any)
United Kingdom	86444	Vodafone, Orange, 3, O2
Brazil	40404	Nextel, TIM
Haiti	40404	Digicel, Voila
Ireland	51210	Vodafone, O2
India	53000	Bharti Airtel, Videocon, Reliance
Indonesia	89887	AXIS, 3, Telkomsel, Indosat, XL Axiata
Italy	4880804	Wind
3424486444	Vodafone
» See SMS short codes for other countries

1. François Chollet‏Verified account @fchollet 7 Oct 2018
  Fun fact: both Kubernetes and Keras have Greek names. Kubernetes means "helmsman" in Greek (same origin as cybernetics), and Keras means "horn" (after the gates of horn and ivory in book 19 of the Odyssey)
  
  2 replies 11 retweets 65 likes
  
  Show this thread
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Tim Scarfe‏ @ecsquendor 6 Oct 2018
  Replying to @fchollet
  
  Microsoft Azure has a great Kubernetes service too! #justsaying
  
  2 retweets 7 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Marouane‏ @___marouane___ 6 Oct 2018
  Replying to @fchollet
  
  Give a try to @PolyaxonAI . A great open source tool for building, training, and monitoring large scale keras applications.
  
  1 retweet 7 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Ash‏ @null_recurring 6 Oct 2018
  Replying to @fchollet
  
  We already have ML in k8s in production btw. One of drawbacks having a high-performant application in an orchestration system is that it can surpass the memory limit in a very short spike (a few ms) and no monitoring system will be able to catch it
  
  2 retweets 2 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Hamza BEN AMARA‏ @Deeper_Hamza 6 Oct 2018
  Replying to @fchollet
  
  I see a tremendous benefits for using containers in AI models but unfortunately large companies won't try to build infrastructure on baremetals with GPU to harness the portability and isolation between dev/prod environments. We're still in the phase of VMs and Cloud IaaS.
  
  1 retweet
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Sandeep Srinivasa‏ @sandeepssrin 6 Oct 2018
  Replying to @fchollet
  
  "The next issue relates to Tensorflow. To be completely honest, I have no idea why this works. But it does. So let’s get this thing running." TF really needs a better UX :(
  
  4 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Times and Tech‏ @TimesandTech 6 Oct 2018
  Replying to @fchollet
  
  Nice
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Holger H‏ @holken 7 Oct 2018
  Replying to @fchollet @ShirinGlander
  
  This works very well with R #rstats as well. Check out the plumber package.
  
  2 likes
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo
1. Ronald Das‏ @DasRonald 8 Oct 2018
  Replying to @fchollet
  
  @Mitra_Abhish
  Thanks. Twitter will use this to make your timeline better. Undo
  
  Undo