PyCon DE & PyData Berlin 2022
– · bcc Berlin Congress Center
The 2022 edition of PyCon DE & PyData Berlin took place 11–13 April 2022 at the bcc Berlin Congress Center. Every session is listed below, with the speakers on their own page. Revisit the original conference site.
-
Making Machine Learning Applications Fast and Simple with ONNX
Taking trained machine learning models from inside a Jupyter notebook and deploying them into a production microservice is painful for two reasons: -…
-
Recording available. pytest - simple, rapid and fun testing with Python (3 hours)
The pytest tool presents a rapid and simple way to write tests for your Python code. This training gives an introduction with exercises to some…
-
Recording available. Efficient data labelling with weak supervision
Data labelling is often considered a separate task that takes place before the real "machine learning work" happens, similar to waterfall software…
-
Honey, I shrunk the target variable! Common pitfalls when transforming the target variable and how to exploit transformations.
Feature engineering takes up a huge part in the work-life of a data scientist. Sometimes this doesn't stop at features but also the target variable itself…
-
Recording available. Web based live visualisation of sensor data
Raspberry Pis are great for interfacing directly with sensor hardware. However, users might prefer their familiar desktop computer or any mobile device to…
-
Recording available. Do I need to be Dr. Frankenstein to create real-ish synthetic data?
Synthetic datasets have caught the fancy of researchers, statisticians, and analysts. Also called fake or proxy data, not only does it address the privacy…
-
Introduction to OPC-UA and industrial IoT: Liberate machines from the proprietary clutches of Big Hardware with the power of opcua-asyncio
Industrial IoT as a field has not outgrown proprietary protocols and closed systems, and this hinders the field in countless ways and funnels money into…
-
Recording available. Machine Learning Testing Ecosystem of Python
In this talk, I'll present the growing ecosystem of machine learning testing tools in Python. Machine learning validation and testing is an emerging…
-
Recording available. Trojan Source Malware - Can we trust open-source anymore?
Recently, a paper is published to demonstrate how a visibly valid contribution can contain malicious code by exporting the Unicode control characters.…
-
Recording available. (Serious) Time for Time Series
From inventory to website visitors, resource planning to financial data, time-series data is all around us. Knowing what comes next is key to success in…
-
Recording available. But this is an OAuth, is it not?
OAuth simplified and secured third-party integrations for the end user. But for the developer of the integration, it can still present some friction. This…
-
Recording available. How to build a Python-based Research Cloud Platform from scratch
This talk will present the journey of a quantitative asset manager from an outdated (non-Python) onPrem research setup to a modern Python-centric cloud…
-
Recording available. Introduction to Uplift Modeling
In this talk we introduce uplift modelling, a method to estimate causal effects of a treatment, e.g. a marketing campaign, to effectively target customers…
-
Recording available. Using a database in a data science project - Lessons learned in production
Storing and processing data in a relational database for a machine learning project presents unique challenges. Processing large volumes can take long,…
-
Recording available. Data Science at Scale with Dask
A Pythonic introduction to methods for scaling your data science and machine learning work to larger datasets and larger models with Dask, all while…
-
Easy and flexible imaging with the Core Imaging Library
In this talk, we present the [Core Imaging Library (CIL)](https://github.com/TomographicImaging/CIL), an open-source, object-oriented Python library for…
-
Recording available. Making MLOps uncool again
Machine learning operations (MLOps) have gained attention among practitioners aiming to automate the development of Machine Learning models, attempting to…
-
Recording available. Data Apis: Standardization of N-dimensional arrays and dataframes
We would like to introduce the consortium of Data APIs, where we will be presenting our motivation, objectives and progress of the standardization process…
-
Recording available. Detecting drift: how to evaluate and explore data drift in machine learning systems
When your ML model is in production, you might observe input data and prediction drift. In absence of ground truth, drift can serve as a proxy for the…
-
Recording available. Impact of Cultivating a Diverse and Inclusive Workplace
Let’s face it. The positive impact of diversity and inclusion is no longer debatable. Although diversity and inclusion (D&I) offers clear benefits, it’s…
-
Recording available. Sankey Plots with Python
This talk provides an introduction to Sankey plots with a focus on creating them with various Python libraries. We will talk about the pros and cons of…
-
Recording available. The secret sauce of data science management
The question - “how to become a successful data scientist?” is often discussed in conferences like these. Today I would like to address a second level…
-
Recording available. "Easy Python": lies, damned lies, and metaclasses
We often refer to Python as a "simple, executable pseudo-code" language. Such statements appear in books, tutorials and all over the Internet. But after…
-
Recording available. Creating 3D Maps using Python
In this talk it is shown how to create 3D Maps using Open Data and Python. There are many open data sources available now for direct download, for example…
-
Recording available. Come as you are: Transitioning from Science to Data Science
I would like to give a little insight into my journey from academia to industry. I started working as a data scientist after more than a decade in…
-
ML Communication 101: How to talk about Machine Learning with anyone
Speaking the same language is already quite challenging when research scientists from academia communicate with data scientists from industry. Still,…
-
Recording available. Rewriting your R analysis code in Python
R and Python are two of the most powerful tools for any kind of data analysis. But both programming languages have their strengths and weaknesses. This…
-
Recording available. Speeding up Python with Zig
Zig is an almost drop-in replacement for C, and that includes directly importing and calling the Python C API without the use of FFI and/or bindings. This…
-
Recording available. Squirrel - Efficient Data Loading for Large-Scale Deep Learning
Data stall in deep learning training refers to the case where combined throughput of data loading and transformation is less than the consumption rate of…
-
Recording available. What I learned from monitoring more than 30 Machine Learning Use Cases
This talk summarizes the most important insights I gained from running more than 30 machine learning use cases in production. We will take a loan…
-
Your data, your insights: creating personal data projects to (re-)own the data you share
We are all constantly collecting and sharing personal data in exchange for services. And although often this data is available to us as users as well,…
-
conda-forge: supporting the growth of the volunteer-driven, community-based packaging project
The conda-forge project is one of the fastest growing Open Source communities out there – and most data scientists have probably heard of it. In this talk…
-
Recording available. Flexible ML Experiment Tracking System for Python Coders with DVC and Streamlit
There are tons of tools to do data science. Too often, data scientists end up using a monolithic AI platform that “does everything by clicking on a UI”.…
-
Recording available. Fundamentals of relational databases
Are you somewhat comfortable with using SQL to access data, but are curious to know what happens behind the scenes when you send off your query? Then this…
-
Recording available. How to Find Your Way Through a Million Lines of Code
No matter whether you are starting a new job or you want to contribute to an open-source project, learning a new code base can be intimidating. This is…
-
Recording available. Inpsect and try to interpret your scikit-learn machine-learning models
This tutorial presents the different inspection techniques currently available to inspect a machine-learning model developed with scikit-learn. In…
-
Recording available. JupyterLite: Jupyter ❤️ WebAssembly ❤️ Python
JupyterLite is a JupyterLab distribution that runs entirely in the web browser, backed by in-browser language kernels such as the WebAssembly powered…
-
Recording available. Make the most of Django
Taking full advantage of **Open Source** software means getting involved in its **community** and **contributing** to its development. We will see how…
-
Recording available. Python 3.10: Welcome to pattern matching!
Python 3.10 is out and a new way of thinking about programming was brought to the Python language - Pattern Matching. We will go through how to use this…
-
Recording available. Refactoring
In this tutorial, you will refactor a space travel text adventure. Starting with a working but messy program, you will improve the structure of the code.…
-
Recording available. Reproducible machine learning and science with python
With machine learning being used in all domains of science, reproducibility and openness is major concern for these experiments and workflows, This…
-
Recording available. Securing Django Applications
Django is the most popular Python-based web framework used for creating web applications. The web applications are vulnerable for various reasons…
-
The Magic of Python Objects
Objects are at the heart of the magic of Python. Their 125+ magic methods hide the complexity of the underlying engine and cover it with a layer of…
-
Recording available. There Are Python 2 Relics in Your Code
Are you done with the migration to Python 3 or did you involuntarily leave some Python 2 relics in your code? Did you leave it as a backdoor so you can…
-
Recording available. XAI meets Natural Language Processing
The idea of Explainable AI (XAI) gets more and more attention, as customers and users want to understand AI models and the reason behind predictions. But…
-
Recording available. You shall not share!
Online social network users frequently share personal information online. While each post is targeted to a certain audience, it is not always easy to…
-
Recording available. Forget Mono vs. Multi-Repo - Building Centralized Git Workflows with Python
The mono vs. multi-repo is an age-old debate in the DevOpsphere, and one that can still cause flame wars. What if I were to tell you that you don't have…
-
Recording available. How to Trust Your Deep Learning Code
Errors in Deep Learning are hard to catch as training often fails silently s. Unit testing can catch those errors early and lets you make changes to your…
-
Recording available. Introducing the Dask Active Memory Manager
The Active Memory Manager is a new experimental feature of Dask which aims to reduce the memory footprint of the cluster, prevent hard to debug out-of-…
-
Demystifying Python's Internals: Diving into CPython by implementing a pipe operator
Diving into the CPython source code can feel daunting. Whether you want to start contributing or just want to get a better understanding of Python by…
-
Recording available. Biases in Language Models
The talk is an attempt to measure biases in most popular language models and we propose a solution to reduce the bias, and promote social inclusion and…
-
Building a Sign-to-Speech prototype with TensorFlow, Pytorch and DeepStack: How it happened & What I learned
Building an E2E working prototype that detects sign language meanings in images/videos and generates equivalent, realistic voice of words communicated by…
-
Recording available. Secure ML: Automated Security Best Practices in Machine Learning
As data science capabilities scale, the core concept of security becomes growingly critical. In this talk we will introduce the security challenges and…
-
Recording available. Unclear Code Hurts
Code may work or not, but it will always tell a story. Computers will not complain about how you write it (except correct syntax), but human readers will.…
-
Grokking LIME: How can we explain why an image classifier "knows" what’s in a photo without looking inside the model?
Many machine learning models are too complex for humans to comprehend. Algorithms like LIME can explain model outputs, even without looking at a model’s…
-
Upgrade your Documentation to the Next Level
Documentation is critical in creating a successful open source project. As open source maintainers we struggle to have a great documentation because of…
-
Recording available. Do we really need Data Scientists?
The term Data Science has emerged in the late 70s in a wider spectrum of research and industry and since then we observed an increasing trend for it,…
-
Recording available. How to deal with toxic people
As an open source maintainer, sooner or later you'll encounter ungrateful, entitled or outright toxic people who can be a real drain on your motivation…
-
Recording available. It is all about files and HTTP
Why do some .pdf files open in a browser at a lick? At the same time, others wait for the answer of where to download? I know that some pages load as fast…
-
Recording available. Battle of Pipelines - who will win python orchestration in 2022?
Python is used for a wide variety of applications in data science, scientific computing, and other complex enterprise-level products. When building…
-
Recording available. The state of DevOps for Python projects
Let's take a look at the DevOps space through the eyes of a team of Python developers. Where does it stand today and where are we going? This talk…
-
Recording available. What are data unit tests and why we need them
I will introduce the concept of data unit tests and why they are important in the workflow of data scientists when building data products. In this talk,…
-
Recording available. 5 Steps to Speed Up Your Data-Analysis on a Single Core
Your data analysis pipeline works. *Nice.*<br/> Could it be faster? *Probably.*<br/> Do you need to parallelize? *Not yet.* We'll go through optimization…
-
Recording available. Fast native data structures: C/C++ from Python
This talk will show how compiling your Python code with Cython (https://cython.org/) enables you to make direct use of fast and memory efficient native…
-
Recording available. Faster Workflow with Testdriven Development
I will show how testdriven development will actually be an asset which can be less time consuming than the default approach for most developers. In this…
-
Recording available. jsonargparse - Say goodbye to configuration hassles
Did you ever find yourself in one of the following situations? * You hard-coded a parameter and now you really wish you could easily change it. * A proper…
-
Recording available. PPML: Machine Learning on Data you cannot see
What if I tell you that you can run a complete ML pipeline on private data, without any anonymisation, nor even accessing the data in the first place? 🧐…
-
Transformer based clustering: Identifying product clusters for E-commerce
In this talk we will be presenting a case-study of how we used two open source libraries, Sentence-Transformers and Facebook Faiss, to successfully…
-
We know what your app did last summer. Do you? Observing Python applications using Prometheus.
When your service is running in production do you know if it’s performing as intended? How can you know if latency for requests is higher than expected,…
-
Recording available. 5 Things You Want to Know About AI Adoption in the Enterprise
All one needs is strategy, skill and resources to make digitalization and AI happen. So why is everything taking so long? Shouldn’t you all be finished…
-
Challenge Accepted - How to Escape the Quicksand While Engineering a Computer Vision Application
Have you ever been in a situation where you do not know the next step? This talk presents the challenges encountered while building a Computer Vision…
-
Recording available. Practical graph neural networks in Python with TensorFlow and Spektral
Graph neural networks (GNNs) have become one of the hottest research topics in recent years. Their popularity is reinforced by hugely successful industry…
-
Quitting pip: How we use git submodules to manage internal dependencies that require fast iteration
Everyone knows the pain of dependency management and things become especially complicated when you have to manage internal dependencies across many…
-
Recording available. `python-m5p` - M5 Prime regression trees in python, compliant with scikit-learn
Regression trees are powerful Machine Learning models capable of both flexibility in modeling as well as interpretability when the tree is not too deep.…
-
Recording available. A data scientist's guide to code reviews
A crucial aspect of software engineering teams' working agreements are code reviews. By applying the four-eyes principle on code, teams can reduce the…
-
Recording available. Advanced Django ORM
The Django ORM is one of the most powerful ORM in the Python space. We will look at implementing complex queries and constraints as well as performance…
-
An Introduction to Inter Process Communication and Synchronization using Python
IPC and Synchronization techniques using Python will be explained in this tutorial. We will learn how to pass messages to processes using Queue and Pipes,…
-
Recording available. Aspect-oriented Programming - Diving deep into Decorators
The aspect-oriented programming paradigm can support the separation of cross-cutting concerns such as logging, caching, or checking of permissions. This…
-
Recording available. Building an ORM from scratch
Have you ever wondered how Django, SQLAlchemy, and PonyORM can translate Python objects into database rows? How complicated can this really be? In this…
-
Can you Read This? (Or: how I Improved Text Readability on the Web for the Visually Impaired)
I will explain my approach of detecting texts on top of an image background that are unreadable to people with visual impairment. I will explain the…
-
Recording available. deepdoctection - An open source package for document intelligence
Extracting information from business documents is difficult. They often have a complex visual structure and the information they contain is not tagged.…
-
Recording available. Easily build interactive plots and apps with hvPlot
Do you use the `.plot()` API of pandas or xarray? Do you ever wish it was easier for you or your collaborators to try out different combinations of the…
-
Financial Portfolio Management with Deep Reinforcement Learning
This talk will answer the question, if a deep reinforcement agent can be trained to optimize a financial portfolio. And therefore outperform an etf-based…
-
Forget ‘web 3.0’, let's talk about ‘web 0.0’. A brief history of the Internet, and the World Wide Web.
For all the excited talk of the future and Web 3.0, it can be easy to overlook the past. How and why did the Internet come to be? What about the World…
-
How a simple streamlit dashboard will help to put your machine learning model in production
Have you struggled getting your valuable machine learning model into the hands of users? No matter how good your ML model is; a lack of monitoring and…
-
Recording available. Introduction to MLOps with MLflow
Machine learning requires experimenting with different datasets, data preparation steps, and algorithms to build a model that maximizes some target…
-
Recording available. Navigating the limitations of Python’s concurrency model in web services
How do web frameworks navigate the limitations of the Python concurrency model? In this talk we will dive into the inner workings of some of Python’s most…
-
Recording available. On Blocks, Copies and Views: updating pandas' internals
a.k.a. “Getting rid of the SettingWithCopyWarning” Pandas’ current behavior on whether indexing returns a view or copy is confusing, even for experienced…
-
Performing Content: Can NLP and Deep Learning algorithms predict reader preferences?
Can we predict a reader's engagement time before publishing an article? This talk presents a use case performed together with a regional German newspaper…
-
Recording available. Predictive Maintenance and Anomaly Detection for Wind Energy
This talk will describe predictive modeling applications in wind turbine maintenance, the challenges of anomaly detection and ways to move to more…
-
Recording available. Processing Open Street Map Data with Python and PostgreSQL
The Open Street Map (OSM) project is a global, open-source database with over 50GB of data, and this number grows everyday with every user submission.…
-
Recording available. Python 3.11 in the Web Browser - A Journey Keynote
Python 3.11 alpha comes with experimental support for Web Assembly and can be built to run in modern web browsers or Node.js out of the box. I’m going to…
-
sktime - python toolbox for time series: advanced forecasting - probabilistic, global and hierarchical
sktime is a widely used scikit-learn compatible library for learning with time series. The forecasting module of sktime provides a unified, sklearn-…
-
Recording available. Slack bots 101: An introduction into slack bot-based workflow automation
Most developers work with Slack every day, yet very few of them know about the awesome things you can do when you build your own slack bot. For example,…
-
Recording available. Stupid Things I've Done With Python
On every computer I've had for the past 20 years, I've created a folder called "stupid python tricks". It's where I put code that should never see the…
-
Recording available. The Myth of Neutrality: How AI is widening social divides
Many people expect artificial intelligence to be neutral - or at least more objective than we humans are. But is it really? In recent years, researchers…
-
Recording available. Unsupervised shallow learning for fraud detection on marketplaces
Combating fraud, scams and wrongdoings in large marketplaces and platforms that connect millions of individuals as sellers and shoppers poses a very…
-
Beyond the basics: Contributor experience, diversity and culture in open source projects Keynote
How can we go beyond the basics when engaging new contributors and improving an open-source project's culture around inclusiveness, accessibility, and…
-
Recording available. 5 Years, 10 Sprints, A scikit-learn Open Source Journey Keynote
We all use open source tools in various capacities, yet knowing how to contribute to open source is not as well known or accessible. The limited knowledge…
-
Recording available. 5 Things we've learned building large APIs with FastAPI
FastAPI has become a very popular framework for building high-performance and well documented APIs in Python 3.6+. In this talk we share 5 things we've…
-
Recording available. Optimize your network inference time with OpenVINO
During the talk, I'll present the OpenVINO™ Toolkit. You'll learn how to automatically convert the model using Model Optimizer and how to run the…
-
Recording available. Career Panel
Working in the 21st century is very different from the past. Especially the digital space offers many opportunities to work in the office and remotely. In…
-
Overcoming 5 Hurdles to Using Jupyter Notebooks for Data Science, by the JetBrains Datalore Team
Jupyter notebooks have pretty much become the standard for data science and data analysis teams. However, there’s still a number of pain points when it…
-
Seeing the needle AND the haystack: single-datapoint selection for billion-point datasets
Python tools like Bokeh and Dash let you build custom Web-based interactive visualization apps and dashboards. While these solutions work well to…
-
A Smooth Ride: Online Car Buying and Selling at mobile.de
Mobile.de is Germany's largest online vehicle marketplace. Every day, thousands of people come to our platform looking to buy or sell a vehicle. Buying or…
-
Recording available. My forecast is better than yours! What does that even mean?
Forecasting is one of the most popular applications of Machine Learning. In the last decades, it went from large numbers, few factors, and simple…
-
Recording available. Python for Everyone - PyLadies' Insights Panel Discussion
PyLadies Mission is to advance a diverse Python Community. This work is done by many volunteers in PyLadies chapters around the world, and here in…