Engineering Blog

2018 is over, and since we are well into 2019 it’s a good time to summarize the state of the multi-cloud business as we have experienced it at DoiT International during 2018, and how we see the future of multi-cloud during 2019 and beyond.

Kaggle Days is (almost) officially the most interesting event to meet, learn and compete against the most talented data scientists worldwide. And this is how we won it.

From the very beginning and since the establishment of DoiT International in late 2010, our vision was to create a global cloud-solution-reseller business while providing our customers with outstanding support on core cloud infrastructure, machine learning, and data analytics.

Sometimes you will need to call the Kubernetes API from outside the cluster whether it’s to run a job or to list deployments, or anything else the Kubernetes API provides.

As part of our technical support operations, we have noticed that our customer’s Tomcat server on Google Cloud Platform is not using the correct remote IP address in the access log.

Building and deploying production-grade machine learning models can be somewhat tricky. Even with technologies like Google Cloud AutoML, Cloud ML Engine and other out-of-the-box machine learning tools, training models and using them in production systems commonly requires a vast set of skills that can include some advanced Python programming, understanding complex models, SQL and DB technologies. This blog post demonstrates how to build a prediction system for shared cars/bikes/scooters using very simple tools!

Most of the time, we scale our Kubernetes deployments based on metrics such as CPU or memory consumption, but sometimes we need to scale based on external metrics. In this post, I’ll guide you through the process of setting up Horizontal Pod Autoscaler (HPA) autoscaling using any Stackdriver metric; specifically we’ll use the Request Per Second from a Google Cloud HTTP/S Load Balancer.

Both XGBoost and TensorFlow are very capable machine learning frameworks but how do you know which one you need? Or perhaps you need both?

Ephi is a very simple bot for Slack that allows users to send ephemeral messages that are automatically deleted for everyone (i.e. self-destruct messages) after a set period of time. Yes, it’s like Snapchat for Slack, but without the nudity, hopefully ;-)

With the massive adoption of Kubernetes and the fact the continuous delivery has become a standard practice, the rollout of new versions is now more automated than ever. But what happens if you deploy a faulty version? How much time and effort it will take to rollback to the previous, non-faulty version?

Recently, I have completed two Google Cloud certifications, - Professional Cloud Architect and Professional Data Engineer. The team at Google Tel-Aviv were nice about it and gave me a Google Home device as a token of their appreciation (thank you Nir Atias & Ifat Yanai!)

TL;DR: Amazon SageMaker offers an unprecedented easy way of implementing machine learning pipelines, significantly shortening the time to market for data scientists and engineers.