Drew Hodun

@googlecloud. All things Data + ML. Brewing beer, mountains, Star Wars. All opinions are my own.

[Drew Hodun @DrewHodun]

So much gratitude for everyone who helped make this project happen.

And of course, thanks to Rui Costa for inviting me along on the journey.

[Vic Iglesias @vicnastea]

Kubernetes friends...

Think of clusters as cells.

Make many of them to reduce the blast radius of changes (upgrades, deployments, tool installations, etc).

It's hard to move to this model after you've already scaled up significantly.

[Drew Hodun @DrewHodun]

@marcelotryle @FastAPI GKE and @skaffolddev for deployment. skaffold lets us automatically build and deploy our services, either locally to minkube or using Cloud Build and GKE. Even allows us to debug all our services running on GKE as if they were running locally!

[Drew Hodun @DrewHodun]

Cattle not pets.

https://t.co/vEllgR74X3

[Drew Hodun @DrewHodun]

@brian_suk Have always managed to just stop myself from pointing this out when someone claims they have so many acronyms. And if you do mention, the answer is always "Oh you don't understand, we have a dictionary/wiki/etc since we have so many acronymns....." 🙄

[Vic Iglesias @vicnastea]

Cant get Docker running on your machine for some reason?

Mount Cloud Shell's Daemon locally with:

gcloud alpha cloud-shell ssh -- -nNT -L `pwd`/docker.sock:/var/run/docker.sock

Then use it in another terminal:
export DOCKER_HOST=unix://docker.sock
docker run hello-world

[Diane Chaleff @DianeChaleff]

Ever had someone make the wrong assumption about your value, your skills, or your leadership? I corrected this with three simple secrets - to leverage the power of my introduction. Thanks allies @GrannisWill and @UlkuRowe for inspiring!
https://t.co/dS6sdoiPPI

[Murph @NegatedVoid]

I don't know where you all get your breaking news alerts from, but me? It's PagerDuty.

[Drew Hodun @DrewHodun]

Vote.

[so called parody. @cloud_opinion]

AWS is now legacy cloud - sorry to break your bubble thought leaders.

[lak lakshmanan @lak_luster]

@Di_Ku The size of language ML models means that many more corner cases can be implicitly memorized, in conjunction with user attrs. I found Docs spell-correcting / suggesting completions on Keras and tensorflow api functions ... It's going to get better due to https://t.co/keIdNS377s

[Corey Quinn @QuinnyPig]

So the recent "Genderify" kerfuffle reminded me of something a couple of years ago.

I was gifted an @awscloud DeepLens and wanted to do something with it for re:Invent (AWS’s own version of Cloud Next).

[Drew Hodun @DrewHodun]

Psyched that Google Cloud continues to get the latest @nvidia GPUs into the hands of our customers fast. A100s are pretty slick!

https://t.co/4kWaCLOHoV

[Jordan Tigani @jrdntgn]

BigQuery goes multi-cloud. From the early days of BQ, we said "move the compute to the data, not the data to the compute". Now you can move your compute to data that lives on other clouds (AWS now, Azure soon), just like it was another region.

[François Chollet @fchollet]

Want to train your Keras models on a beefy machine on GCP (or just many machines)? With TensorFlow Cloud, you can just add a single line to your script or notebook and get it running -- no config work necessary past the initial setup.

Try it: https://t.co/49NPl0YFvs

[Felipe Hoffa @felipehoffa]

I just published "Stack Overflow future trends: Predicting with ARIMA and BigQuery"

https://t.co/KwAy8J8DIF

Can you predict the top @StackOverflow tags of 2023? Sure! Get your time-series analysis started with SQL now.

Plus a Data Studio dashboard and a @getdbt pipeline.

[Drew Hodun @DrewHodun]

Wish you had fancier Gmail filters that could use regex?

[James Brown @ibjhb]

This is an open-sourced version of a Gmail labeling tool I use internally at Google and I STRONGLY recommend it. It was written by @ranmocy

I just set it up on my personal account:
https://t.co/D6kFZpnIhP

[Khalil @sehnaoui]

The sound of Jello being shredded is definately not what I was expecting ...

[Drew Hodun @DrewHodun]

I just published in @TDataScience How to extend a Keras Model https://t.co/jh3kMo1cYa

Came across this with a number of customers doing large-scale ML on @googlecloud, figured I'd write it up. Thanks to @lak_gcp for the help!

[Martin Görner @martin_gorner]

The classic "Tensorflow without a PhD" tutorial has been updated for Keras and Tensorflow 2.2: https://t.co/SuLIlwwHQG

After many years, still a good place to start learning about neural networks, relu, softmax, cross-entropy, learning rates, dropout, batch norm, ...
Enjoy!