The first day of EuroPython has been nothing but fun. But was it really the first day… For some of us, it was. Two team members arrived yesterday, while one of us already got here on Sunday. Perfectly on time to attend the training sessions.

For the booth staff, today was all about getting to know the attendees. Find out what they are working on right now, discuss developments in the world of Python and occasionally talk about Lego. Lego? Yes, Lego.

We dare to say that we have found the coolest giveaway that fits our product perfectly. It’s the Lego container yard. And of course the famous Dutch stroopwafels. So if you haven’t stopped by our booth to have your badge scanned and grab a cookie, come to find us tomorrow. But, in between scanning the badges of potential winners, we also give a live demo about deploying code and applications on our container engine that is based on Kubernetes and OpenShift origin. We will give another demo on Thursday and Friday during the lunchbreaks.

Team member Freark is in charge of visiting the training and talks this week. Today he shares his notes on the following sessions:

Opening keynote “Die Threads” by David Beazley

As always David Beazley is a very engaging talker. It’s amazing to see him be able to live code and talks at the same time. In this presentation, he was talking about and coding with a (little threading library)[https://github.com/dabeaz/thredo] he wrote as an experiment to make working with threads easier. Among the problems which are solved with this library is the problem of properly terminating a threaded program with ctrl-c. This library is using curio and thus asyncio too. In a more philosophical sidetrack of the talk, the audience is asked: “When thread sleep, do they dream?”

“Understanding and Implementing Recurrent Neural Networks using Python” by Anmol Krishan Sachdeva

In this presentation, Anmol is briefly explaining the concept of recurrent neural networks and in particular LSTM network. As a use-case he presented a Python notebook for predicting stock prices and took us through steps like normalizing your input data, building your neural network layers using the Keras library and showed the results of the training. But before you go and invest all your savings in the stock market, Anmol also shared the wise statement “All models are wrong, but some are useful”.

“Hacking Reinforcement Learning” by Guillem Duran

In this presentation, Guillem shared with the audience endeavors to develop a variation of machine learning dubbed FractalAI. To be honest, I did not fully understand what the nuanced differences were, but he did show pretty nice demos where he built AI for various old-time video games trained with this approach which apparently took much less learning time than most other deep learning approaches. Something I should probably play with a bit in the future to understand better.

Other talks

Besides the before mentioned talks, learning, in a fast-paced speed, about Trio as a more pythonic way of using asyncio from Emmanuel Leblond was interesting. Just as fast paced was Stephane Wirtel’s talk “What’s new in Python 3.7″ giving an overview of improvements and new features in the latest and greatest Python version.

We look forward to the second day at EuroPython where we can discover, learn and share knowledge with the Python community! What are your key takeaways of the first day of EuroPython?