EuroPython2022

Overview

This is an article to share my wonderful experience participating in the EuroPython Conference in Dublin, Ireland.

Objective

• I'd like to share the key points to help my fellow Python Developers (Pythonistas).
• I'd like to keep a journal of my learning. In one week, I believe I've improved my programming skills to write optimized faster Python code, use modern tools that make our jobs easier, and my social skills to the highest level, as I've made some new friends or acquaintances from all over Europe (Germany, Sweden, Amsterdam, France, Italy, Spain, Ireland, UK, Portugal).
• The consciousness around Artificial Intelligence.

Tutorials

The first two days of the conference were dedicated to the intensive hands-on training on various Python programming language topics. The ones I liked are listed below.

Python under the Hood

I'm sure we've all used the __init__ constructor method in Python. Some of us are experienced programmers who have used the __str__ and __repr__ methods. But do you know the function constructor method for Immutable Float or Tuple objects? It's the __new__ dunder/magic method. Try to override and create a new object.

Do you always raise NotImplemented Error? Have you tried returning NotImplemented, which goes to the Reflection function and raises the appropriate error if it is not found? Is this your first time hearing it? Check on Stackoverflow.

This training by Rodrigo Gir?o Serr?o has been my favorite. This has been possible because of the way he teaches. Check out his Lightning talk and you will understand he's the best.

Test Driven Development

To develop unit tests, one of the training was more focussed on property-based testing using the Hypothesis library. Instead of setting up test cases by developers, it generates the test data based on the given specification and performs the tests.

The other training was more driven by understanding the TDD style, and the basics of creating unit tests using Pytest.

Building APIs with GraphQL

I took a basic first session of GraphQL and my understanding of GraphQL is that it helps in exposing all use case functions under one endpoint and only uses the 'POST' of the HTTP protocol for the CRUD features, and authentication.

Keynotes

Dodging AI Dystopia By Dr. Nakeema Stefflbauer, EMBA

The key takeaways from this talk,

AI solutions looking for problems - A solution looking for a problem, instead of a problem looking for a solution. :(

• The targeted adverts are produced using machine learning algorithms. Analyzing digital payment history and grouping users into several categories, such as a sandwich, high spenders, and affluent people. Influencing additional purchases from consumers. This worries me and prevents me from making digital payments. I'll be bombarded with advertisements to buy XYZ. The same is true with search results.
• If Algorithms can be more trusted than humans in law enforcement. It depends on the accuracy and how AI systems are trained, the sample data used to train them, if accuracy is not good and the system may introduce biases, maybe just like humans do.
• If the conclusions of an algorithm are used to guide hiring decisions. Candidates spend more time making changes to their resumes to make them stand out in ATSs powered by ML. Keep in mind that Amazon pulled down the platform that penalized women and gave them lower rankings.

Killer Robots Considered Harmful aka AI Weapons by Laura Nolan

• The purpose of this lecture was to raise awareness about the morality of nations competing with one another to develop automatic systems that can choose targets and launch attacks without human intervention.

My key takeaway from both of these AI sessions is we should focus on using Artificial Intelligence to solve problems, but we are diverted and distracted to create more problems using AI solutions. It shouldn't be like this, a solution for one, a problem for another. We need to counter common problems like public healthcare, climate control, and others. There are many ways AI can be used for GOOD. Ethics in AI is an essential subject for everyone in this field.

Multithreaded Python by Sam Gross

• This was another great session to understand the limitations of python in executing the code in parallel, the team has created a 'nogil' project soon to merge upstream to run multithreaded python code efficiently https://github.com/colesbury/nogil.
• I'm going to play around with it to try distributed computing.

Talks

Writing Faster Python 3 by Sebastian Witowski

• Did you know list comprehensions are faster than for loops?
• Did you know generator expressions reduce memory usage?
• Instead of re-using global variables, we should use local variables and return the values needed.
• I encourage everyone to watch the session online on youtube, or at least view the slides.

Python was made for faster development, and Python is slow for people who don't know python well.

Code with Music

In a fun way, he described and I agree a programming language cannot be learned in 30 minutes of a youtube lecture, a week of BootCamp, or a few months of on and off learning. It requires a sheer amount of hard work and real-world experience for a couple of years to completely understand a programming language and build logic on it for a highly scaled production system.

Thank you Nicholas H.Tollervey for this amazing session.

There were a few more sessions I enjoyed and gained knowledge about the Clean Architecture and Poetry tool as a package manager.

Sponsors Booths

It was amazing to interact with Tim Gilboy from Sourcery , I have started exploring their tool that is helping in refactoring bits and pieces of the codebase.

Loved meeting people and getting goodies from Hudson River Trading , Optiver , Workday , and Anvil (Automatically writes python backend for applications while user just needs to place UI elements from their interface.)