Machine learning and computer vision, have been studied for decades, but only recently have begun creeping into industry. This means we need to start applying engineering principles to make models useful, scalable, and portable. This is an introduction into how to continuously train and deploy ML.

Machine learning and computer vision are no longer futuristic concepts or confined to a lab in some university. The industry is rapidly employing ML in products. However, implementing useful models is incredibly difficult because we don’t yet have many tried and tested standards or practices for applying traditional software engineering tools and methods. In this talk, we will dive into how to automate continuous delivery and training of models. We will discuss an example architecture set up for computer vision models, using KubeFlow and Kubernetes. This is intended to be an introductory talk into what is needed when deploying machine learning models, rather than focusing too much on specific technologies.

Anyone involved in data analysis in Python knows NumPy, which is now a de-facto standard for all kinds of data processing in this language. However, it turns out that NumPy lacks many of the recent advancements that other scientific computing packages can provide — take accelerator (GPU) support, automatic differentiation or distributed computing utilities to name a few. PyTorch tries to fill that gap, by providing flexible and Pythonic abstractions for those concepts, while building upon the rest of the Python ecosystem. What’s more, there’s a whole community around the package, which has been steadily growing since its release in 2017, making it the most popular library used for machine learning research these days. This talk is meant to introduce the library while building on familiar concepts from NumPy, and show how PyTorch can be leveraged to express many interesting machine learning and HPC workloads

Connections are a part of us, of the world. From the connections between people, between cultures, within language, and more. In these days when more data is collected daily than we could ever hope to explore, the variety in connections being gathered is opening up the possibility to visualize these (often complex) networks.

During this talk, Nadieh will take you through the design process of several of her (interactive) data visualization works, from personal projects to client work. The common thread they all share is that they all reveal connections, but all differently. From a family tree of 3000 people connected to the European Royal houses to those existing between our Intangible Cultural Heritage created for UNESCO, to connections we have drawn in the night skies and more. Revealing that all types connections are unique and revealing the intricacies that lie within them requires a creative, iterative and custom approach.

Development has changed over the years, from doing everything yourself to a 3rd party package for every function. Operations has changed too, running your own servers is now considered an exception. To the cloud! We have learned that we need to trust others, but as our parents used to say - don’t trust strangers. So we secure our production server more than ever.

Yet, in the middle sits this no man's land: “the build server”. We think it’s time to take a closer look at some of the good practices around securing builds & artifacts to improve our day to day level of trust.

With Marked Sherman statement “Development is now assembly” in mind, the talk will focus more on the package/artifact/repository aspect. Less on the app security inside the code itself or at the OS/Machine level.

This talk I will go into detail on:

• How to verify trust of your dependencies: from metadata, binaries and repositories
• How to provide trust to others that build upon your software
• How this ties into the concept of “reproducible builds”
• How a practical “Software Bill of Material” looks
• How the concepts of the “The Update Framework” (TUF) relate
• How you can implement secure packaging policies

It will explain these topics using practical/code examples from the Nodejs and Docker ecosystems. All this will be presented from the different viewpoints from “dev”, “sec” and “ops”.

Let’s take ownership of your trust, we are already responsible when things go wrong anyway.

Erlang is in many ways quite old though many of the problems for which it used are quite modern. The Erlang language and system was designed around a set of requirements for telecom systems. They were distributed, massively concurrent systems which had to scale with demand, be capable of handling massive peak loads and never fail. The Erlang concurrency and error-handling model was developed around these requirements.

This talk will describe the development of the language and the design of systems based on the Erlang, look at the BEAM which is the Erlang VM, and how it supports the properties of the Erlang language and systems built on top of it. It will also look at further development with the introduction of new languages like Elixir and how they can be added into the Erlang environment - the Erlang ecosystem.

Developers are essential to driving digital transformation on many businesses and this is why executives must learn more about technology and developers should learn more about business. 

During this talk we'll discuss 5 key points about Open Innovation and the Bussiness Platform Model:

 What is the Business Platform Model and how it works

• What is open innovation and how open could it be? 
• What are the most common launch strategies for platforms?
• Governance and policies for business platforms
• Why and How to bring open innovation to your business.