List of videos

Django queries optimization [EuroPython 2018 - Talk - 2018-07-26 - Kilsyth] [Edinburgh, UK] By Ivaylo Donchev Collecting data from multiple Models is a common flow in Django development. In most of the cases prefetch_related and select_related do all of the job for optimizing the queries. When the models structure become complicated enough and we've put our logic for collecting these items in models' properties we suddenly cannot use prefetch_related or select_related anymore. Our View 's queries count depends on selected items count with high complexity. The problem is in the selection logic: we've implemented an algorithm which calculates something and we cannot prefetch or select all of the objects. So this talk's purpose is: To show how to use Django ORM to move the selection logic, the calculations, the aggregations over group of objects from our python code to our database and moving all of the logic for this in one place - the queryset and test it easily. License: This video is licensed under the CC BY-NC-SA 3.0 license: https://creativecommons.org/licenses/by-nc-sa/3.0/ Please see our speaker release agreement for details: https://ep2018.europython.eu/en/speaker-release-agreement/

Standardize Testing in Python [EuroPython 2018 - Talk - 2018-07-26 - Kilsyth] [Edinburgh, UK] By Bernat Gabor In this talk, Bernat will introduce tox, an open source tool with the bold vision of standardizing testing. From a CIs point of view, testing contains much more than just unit and/or integration tests; other things like code style checks, packaging, testing under various versions of Python, and checking that documentation still generates, are just as important. Things quickly start to spiral out of control once you add into the mix that there are many tools and ways to accomplish each of these tasks. In this talk, we will learn how tox tries to abstract away all this complexity, how to easily run CI tests on your local machine, and how one can use this tool inside your CI frameworks to ensure high quality and easily maintainable packages. License: This video is licensed under the CC BY-NC-SA 3.0 license: https://creativecommons.org/licenses/by-nc-sa/3.0/ Please see our speaker release agreement for details: https://ep2018.europython.eu/en/speaker-release-agreement/

When to use Machine Learning: Tips, Tricks and Warnings [EuroPython 2018 - Talk - 2018-07-26 - Fintry [PyData]] [Edinburgh, UK] By Pascal van Kooten Artificial Intelligence, and machine learning in particular, is one of the hottest topics in tech/business. I will explain the core of machine learning, and the main goal of this talk will be to help you judge the success whenever someone yells "I know! let's solve this using machine learning!". I will also provide tips and tricks on how to increase the success of such projects. The second part of the talk will be about 2 open-source python projects I've created, as well as a project I'm working on regarding the trading of cryptocurrency... and their relation to machine learning. Specifically, the challenges and findings in making these cases work will be explored. License: This video is licensed under the CC BY-NC-SA 3.0 license: https://creativecommons.org/licenses/by-nc-sa/3.0/ Please see our speaker release agreement for details: https://ep2018.europython.eu/en/speaker-release-agreement/

Succinct data structures for python [EuroPython 2018 - Talk - 2018-07-26 - Fintry [PyData]] [Edinburgh, UK] By Konstantin Ignatov This is a presentation of and call for participation in development and testing of Python bindings to Succinct Data Structure Library. From Wikipedia: Succinct data structures can represent an object (such as a bitvector or a tree) in space close to the information-theoretic lower bound of the object while supporting operations of the original object efficiently. The theoretical time complexity of an operation performed on the classical data structure and the equivalent succinct data structure are (most of the time) identical. Currently bindings are provided for: Mutable bit-compressed vectors Immutable compressed integer vectors Immutable compressed bit (boolean) vectors Rank and select operations on bitvectors Wavelet trees Comressed suffix arrays Original library: https://github.com/simongog/sdsl-lite Most of examples from SDSL cheat sheet and SDSL tutorial are implemented. License: This video is licensed under the CC BY-NC-SA 3.0 license: https://creativecommons.org/licenses/by-nc-sa/3.0/ Please see our speaker release agreement for details: https://ep2018.europython.eu/en/speaker-release-agreement/

Debugging Your Code with Data Visualization [EuroPython 2018 - Talk - 2018-07-26 - Fintry [PyData]] [Edinburgh, UK] By James Saryerwinnie Let's face it. Sometimes our code just isn't working how we expect it to work. When this happens, we fall back to our trusty tools to help us debug: pdb, the logging module, or even simple print statements. But sometimes that just isn't enough, we still can't figure out why our code is broken. We need something more. There are a number of fantastic libraries in python for creating data visualizations. These libraries are commonly used for analyzing and visualizing large sets of data, but don't see as much usage when it comes to helping us write and debug our code. What if we could combine the two? What if we could leverage data visualization tools to understand what our program is doing? Sometimes, by seeing a visual representation of what our code is doing, it becomes much easier to understand why our code is not working. We don't need to create publication quality graphs and charts, we just need to generate quick, one-off visualizations to understand what our code is doing. What we want is the "data visualization" equivalent to print() statements. To help illustrate this point, I will walk through some of the hardest bugs I've had to track down while working on projects such as the AWS CLI and boto3, the AWS SDK for Python. For each bug, I'll show you how I was able to leverage data visualization techniques to troubleshoot and fix these bugs. Come learn how to debug more efficiently by leveraging data visualization. License: This video is licensed under the CC BY-NC-SA 3.0 license: https://creativecommons.org/licenses/by-nc-sa/3.0/ Please see our speaker release agreement for details: https://ep2018.europython.eu/en/speaker-release-agreement/

May the Fuzz be with you [EuroPython 2018 - Talk - 2018-07-26 - Fintry [PyData]] [Edinburgh, UK] By Heidi Thorpe This talk will explore how Long short-term memory (LSTM) recurrent neural networks combined with Python can use Artificial Intelligence to provide invalid, unexpected, or random data as inputs for testing software and finding network security vulnerabilities. The technique of machine generated test inputs is an established testing approach and is sometimes referred to as fuzzing. I will take this idea further by presenting how various types of machine learning can provide novel outlier test cases to harden software and networks in a way that is superior to using only randomness. Instead, I will discuss how to train neural networks to produce better fuzzing data.This is a natural progression from my work with machine learning and image processing using support vector machines (SVM) and Generative Adversarial Networks (GAN) which I have applied to this new domain of software testing. This a beginner level talk aimed at all software developers and testers with a goal to giving attendees a basic understanding of fuzz testing. Timing of presentation 1 Introduction 0-5 minutes 2 What is a neural fuzzing 5-10 minutes 3 How does fuzzing work 10-15 minutes 4 Steps to train a deep neural network 15-20 minutes 5 Implementing a basic LSTM neural network for security testing 20-25 minutes 6 Conclusion and Questions 25-30 minutes License: This video is licensed under the CC BY-NC-SA 3.0 license: https://creativecommons.org/licenses/by-nc-sa/3.0/ Please see our speaker release agreement for details: https://ep2018.europython.eu/en/speaker-release-agreement/

The rise of Python in the data communities [EuroPython 2018 - Talk - 2018-07-26 - Fintry [PyData]] [Edinburgh, UK] By Alexys Jacob A retrospective and prospective of Python's adoption in the data-driven industries and how it has and should influence its ecosystem and communities. Thanks to its versatility, Python's usage and adoption has changed a lot over the last decade to go beyond the very act of software programming. From Developers to SysOps, closely followed by Scientists and Data analysts, Python has spread to become a common tongue for a wide range of people. We will start by looking at how this increased adoption impacted Python ecosystem and is still shaping it today. While this talk is not walk through all the Python technologies around data, some of them will be outlined so you will hear words like Numpy, Pandas or Jupyter. Then we will try to project ourselves in the future and by highlighting the pitfalls Python has to overcome to keep up with its pace and mature in its ability to scale! Draft of the agenda The rise and collusion of science and engineering and their influence on Python From DevOps to DataOps, the shape and breakthroughs of Python's ecosystem Prospect of challenges and pitfalls in the massive data era License: This video is licensed under the CC BY-NC-SA 3.0 license: https://creativecommons.org/licenses/by-nc-sa/3.0/ Please see our speaker release agreement for details: https://ep2018.europython.eu/en/speaker-release-agreement/

Deep Learning with PyTorch for Fun and Profit [EuroPython 2018 - Talk - 2018-07-26 - Fintry [PyData]] [Edinburgh, UK] By Alexander Hendorf There are all these great blog posts about Deep Learning describing all that awesome stuff. - Is it all that easy? Let's check! We'll look into: style transfer (making a picture look like painting), speech generation (like Siri or Alexa) and text generation (writing a story). In this talk I'll describe the whole journey: A fun ride from the idea to the very end including all the struggles, failures and successes. Steps, we'll cover: The data challenge: get the data ready Have it run on your Mac with PyTorch and an eGPU Creating a character-level language models with an Recurrent Neural Network Creating a text generator Creating artwork License: This video is licensed under the CC BY-NC-SA 3.0 license: https://creativecommons.org/licenses/by-nc-sa/3.0/ Please see our speaker release agreement for details: https://ep2018.europython.eu/en/speaker-release-agreement/

Industrial Machine Learning Pipelines with Python & Airflow [EuroPython 2018 - Talk - 2018-07-26 - Fintry [PyData]] [Edinburgh, UK] By Alejandro Saucedo Industrial Machine Learning This talk will provide key insights on the learnings I have obtained throughout my career building & deploying machine learning systems in critical environments across several sectors. I will provide a deep dive on how to build scalable and distributed machine learning data pipelines using Airflow with a Celery backend. I will also compare Airflow with other technologies available out there and how it differentiates, such as Luigi, Chronos, Pinball, etc. If you attend the talk, you will obtain an understanding on the solid fundamentals of Airflow, together with its caveats and walk-arounds for more complex use-cases. As we proceed with the examples, I will cover the challenges that you will run into when scaling Machine Learning systems, and how Airflow can be used to address these using a manager-worker-queue architecture for distributed processing with Celery. By the end of this talk you will have the knowledge required to build your own industry-ready machine learning pipelines to process data at scale, and I will provide further reading resources so people are able to implement the knowledge obtained almost right away. License: This video is licensed under the CC BY-NC-SA 3.0 license: https://creativecommons.org/licenses/by-nc-sa/3.0/ Please see our speaker release agreement for details: https://ep2018.europython.eu/en/speaker-release-agreement/