Kubernetes and security.
Topic 1: To Helm or Not to Helm? - by Alex Khaerov (25 min)
Topic 2: What you need to know about Kubernetes Security - by Hunter Nield (45 min)
Topic 2 add on: Kubernetes security at Swatmobile - by Vincent De Smet (15 min)
Topic 3: Machine Learning at scale with Kubernetes - by Wesley Goi (30 min)
* To Helm or Not to Helm?
Summary: You probably heard about the first package manager for k8s – Helm. Some already started using it to run and manage applications, despite the recent criticism around this tool.
Does Helm allow you to think in application lifecycle paradigm?
Would ‘homebrew renderer’ really be a better alternative?
Is there a federation angle and kubeadm support to be aware of?
Is Tiller a huge security black hole and how could someone abuse it? We will answer these questions in our talk.
* What you need to know about Kubernetes Security + add-on
Summary: User surveys conducted by Cloud Native Foundation, DataDog & The New Stack indicate Kubernetes is the leading container orchestration framework across on-prem and cloud deployments.
One of the major considerations mentioned by survey respondents is the need for secure deployments. In a distributed, highly scalable environment, typical security patterns will not suffice. In this presentation we will look at security in the context of Kubernetes in terms of containers, deployment and network security.
Provide references on how to ensure a safe journey while adopting these technologies.
The add-on by Vincent will give some practical examples and considerations from adopting some of these requirements at Swatmobile
* Machine Learning at scale with Kubernetes
The need to search through hyperparameter space when tuning a machine learning is an important step in successfully training machine learning. Similarly important is the determination of optimal architectures for Deep Neural Networks. This problem is further exacerbated when the training procedure is multistep.
Data Scientist coming from academia may not be familiar with the constraints of Enterprise compute resource where they previously had access to swathes of raw compute power in the form of HPCs
In his talk, Wesley will share the how his team utilises Kubernetes, to orchestrate ML pipelines and and hyperparameter search.
Alex Khaerov is a development lead at Chainstack, a company providing the ultimate blockchain control panel for business. With 9+ years in web services development, he took on senior software engineer and technical management roles. His main passion is building robust, high-load and distributed service platforms on top of a multi-cloud environment, utilizing all the power of Kubernetes. An avid activist of Moscow Kubernetes and Moscow Python communities. In his spare time, he speaks publicly at tech conferences and secretly codes in Python. Huge fan of laptop stickers.
Hunter Nield is a co-organiser for the Kubernetes Singapore meetups
Vincent De Smet is a co-organiser for the Kubernetes Singapore meetups and works for Swatmobile.io managing the cloud infrastructure.
Wesley is currently working as a Data Scientist at Honestbee on recommendation systems where he develops, deploys and maintains ML services as a µservice. Previously he worked on computational biology in the field of genomics