Banks need to check their customers for money laundering via their transactions and information from other sources (e.g., front desk). Ex-post monitoring of customers is largely done using rule-based approaches, which have a large false-positive rate and are inflexible to detect more advanced and emerging money laundering schemes. Artificial Intelligence (AI) techniques can be used to augment current compliance processes to increase the effectiveness and efficiency of anti-money laundering (AML) activities. We will showcase the potential of AI in AML by presenting a use case, where we utilized unsupervised Machine Learning algorithms to perform KYC consistency checks. We will further highlight, which steps are necessary to cover the full Machine Learning life cycle, from the initial ideation phase, over the Proof of Concept stage, to deploying a model to production, and discuss the technology stack (e.g., Python, PySpark) used for this project.

When we take a look under the hood of modern networks a developer might immediately see similarities to modern software. A network, in essence, is a complex system requiring constant change that needs to be designed, tested, and rolled out in a reliable way. See the similarities to software? In this talk we are going to take a look at how python and open source modules can be used to automate the design, testing and roll-out of a modern network.

In my talk, I will primarily focus on tools for data analysis and data visualization tools that Python offers and also will mention where are current limits of Python in Data Visualization & Analysis and how to overcome them.

It will be a short overview of PyViz and the whole Data Visualization Landscape that Python offers (in each stage of analysis the best libraries will be shown for the specific purpose; as for data visualization we will focus particularly on matplotlib, Seaborn, Plotly but also will mention altair, Bokeh, Datashader, GeoViews, HoloViews, Param, etc.).

In my talk (if there will be enough time) I will also explain how data visualization works (on a high level and on the back-end), what are the most common problems and pitfalls in Data Visualization, and how to avoid it.

In this context, my talk will be also a short "philosophical" introduction to The Grammar of Graphics by Leland Wilkinson and I will also mention which Python libraries currently & already adapted to this vision/philosophy in Computer Science and Data Visualization in general. Also, what are the differences with some “Python native” grammar of graphics like f.e. Vega (altair).

At the end of my talk (if time will allow it) I will also explain what are current trends in Data Visualization (not only in Python) and how to effectively (and possibly) merge the world of BI and AI together (soon/maybe not so far as you might think…).

When you start a new project, typically, the choice between a big framework like Django and a microframework like Flask comes up. This talk will show you that the difference is not as big as one might think and that Django is suitable for projects of all sizes.

At 77k lines of code, Django is a heavy-weight among Python web frameworks. Django comes with lots of batteries included: it has its own ORM, a custom template language, miscellaneous middlewares, and other stuff that sounds not only quite a handful but also a bit dated at a time where everyone is talking about microservice architecture. It is easy to conclude that Django is too complicated, especially for beginners, who might be better served learning a micro-framework.

This talk will show you that this conclusion might be misinformed and that Django is actually a great choice for beginners. We will look at a minimal Django app that is as nimble as an equivalent written in Flask, and you will learn how to pick out the juicy bits from the smorgasbord of features that Django provides.