The talks recordings will be available on our Youtube within 3 months - stay tuned!
Online edition of the annual
Python Conference in Minsk, Belarus
March 13, 2021

Speakers - Python tracks

@mariatta, Canada

Staff Software Engineer
@ambv, Poland

Python core developer
@vitalik, Ukraine

Open-source contributor
Introducing Django Ninja (RU)

The talk is in Russian
@EmeliDral, Russia

Co-founder & CTO
@Evidently AI
@mingrammer, South Korea

Software Engineer
@Karrot Market
@ianozsvald, UK

Principal Data Scientist
@ Mor Consulting
@moshezadka, US

Senior Systems Engineer
@AI_Kho_, Belarus

Principal ML Engineer/Data Scientist
@IHS Markit

Speakers - Data Science track

Speakers - Fireside chats
@sobolevn, Russia

The sessions' MC
@lensvol, Netherlands

Software Developer
@shurph, Belarus

Head of Python+DS/ML Group
Joel Grus is a principal engineer at Capital Group, where he leads a team focused on machine learning and search relevance. Previously he worked as a research engineer at the Allen Institute for AI, a software engineer at Google, and a data scientist at a variety of startups.

He is the author of Data Science from Scratch and Ten Essays on Fizz Buzz, but he may be best known for not liking Jupyter notebooks. He lives in Seattle.
Recently I got sort of addicted to a simple-not-so-simple word game on my phone.
In this talk I'll discuss how I used Python first to cheat then to more deeply explore the space of puzzles and solutions.
Anthony Shaw is a Python advocate from Sydney, Australia. Anthony is a contributor to many open-source communities. Running and contributing to a number of popular open-source tools for DevOps, Security, Automation and Code Quality.

Anthony has been recognized for his contribution to open-source, including a Fellow of the Python Software Foundation and a member of the Apache Software Foundation.

Anthony runs a Python blog and YouTube channel and has recently published a book on the Python compiler.
In this talk you'll see an update to the Pyjion project, a JIT compiler for CPython byte-code.

This project was started 5 years ago but stopped after making no gains in performance. Recent changes to CPython have made optimisations more viable, so now it has been restarted and is showing big performance gains vs. standard CPython with 100% compatibility. Many attempts have been made to JIT compile Python in the fast and few have succeeded.

Is it worth it and what are the gains to be made? This talk will cover the design ideas of a JIT for CPython, optimisations, and future potential.
I'm tiangolo (Sebastián Ramírez), the creator of FastAPI, Typer, and other open-source tools.

I'm currently a developer at Explosion in Berlin, Germany.
If you have used CLI (Command Line Interface) applications, you are probably familiar with shell completion (TAB completion) and how much it helps.

In this talk, you'll learn how to quickly create CLI apps that have automatic shell completion with Typer (FastAPI's little sibling), by declaring standard Python type annotations in your functions.

Typer will do the heavy lifting of handling shell completion for all the major shells, auto-documentation of your CLI, and data validation based on your type annotations.

You can start with just a simple script for yourself, in minutes. And also grow as big as you need, to create complex CLI apps with multiple levels of sub-command groups and many other features.

All by writing minimal code, based on standard type annotations, with autocompletion in your editor, and an intuitive design based on FastAPI.
Mariatta is a Python Core Developer, Staff Software Engineer at Uplight, and the Vancouver PyLadies co-organizer, and one of the founding members of the PyCascades conference.

She moved to Canada almost two decades ago, and now lives in Vancouver with her husband and two children. In her free time, she contributes to open source, builds GitHub bots, and fixes typos.
I consider myself relatively new to the open-source world; my first open-source contribution was in 2016. Since then, I've continued to actively contribute to open source and specifically to core Python and Python libraries.

Pretty soon I found myself being given commit rights to other people's open-source projects. It's been quite a journey. Being a new open-source contributor has its own challenges, and being a new open-source maintainer brings another set of unique challenges.

In this talk, I will share my journey and the things I've learned along the way, and some advice for other aspiring open source maintainers and contributors.
Matteo is a Senior Scientist in the Gray Systems Lab (GSL) at Microsoft, working on scalable Machine Learning systems. Before Microsoft, he was a Postdoctoral Scholar in the CS Department at the University of California, Los Angeles, working on Big Data systems.

Prior to joining UCLA, he was a researcher at the Qatar Computing Research Institute, and at the Institute for Human and Machine Cognition.
He obtained his PhD in Computer Science from the University of Modena and Reggio Emilia.
Karla is a Senior Research Software Development Engineer in the Gray Systems Lab (GSL) at Microsoft. She finished her PhD in Computer Science at the University of Maryland, College Park in 2015.

After graduating, Karla spent 3 years as a research scientist at Intel Labs and joined Microsoft in 2018.
Her research interests are broad, and generally enjoys scalability and performance challenges related to systems infrastructure.
Machine Learning (ML) adoption in the enterprise requires simpler and more efficient software infrastructure. Model scoring, the process of obtaining prediction from a trained model over new data, is a primary contributor to infrastructure complexity and cost, as models are trained once but used many times.

Hummingbird is a novel approach to model scoring, which compiles featurization operators and traditional ML models (e.g., decision trees) into a small set of tensor operations. This approach inherently reduces infrastructure complexity and directly leverages existing investments in Neural Networks' compilers and runtimes to generate efficient computations for both CPU and hardware accelerators.

Hummingbird performance are competitive and even outperforms hand-crafted kernels while enabling seamless end-to-end acceleration of ML pipelines.

Hummingbird is open source, part of the PyTorch Ecosystem, and someone even mentioned it as one of the top 10 Python libraries of 2020!
Machine Learning (ML) adoption in the enterprise requires simpler and more efficient software infrastructure. Model scoring, the process of obtaining prediction from a trained model over new data, is a primary contributor to infrastructure complexity and cost, as models are trained once but used many times.

Hummingbird is a novel approach to model scoring, which compiles featurization operators and traditional ML models (e.g., decision trees) into a small set of tensor operations. This approach inherently reduces infrastructure complexity and directly leverages existing investments in Neural Networks' compilers and runtimes to generate efficient computations for both CPU and hardware accelerators.

Hummingbird performance are competitive and even outperforms hand-crafted kernels while enabling seamless end-to-end acceleration of ML pipelines.

Hummingbird is open source, part of the PyTorch Ecosystem, and someone even mentioned it as one of the top 10 Python libraries of 2020!
I have been working in Machine learning and Data science for several years in a few different IT companies. In particular, I enjoy exploring interesting world problems and solving them with state-of-the-art techniques.

I have developed several open-source python packages, I am the core-contributor of `pytorch-lightning` and actively participating in other well-known projects.
Gene is an artist and a programmer who is interested in autonomous systems, collective intelligence, generative art, and computer science. He is a collaborator within numerous open-source software projects, and gives workshops and lectures on topics at the intersection of code and art.

Gene initiated ml4a, a free book about machine learning for creative practice, and regularly publishes video lectures, writings, and tutorials to facilitate a greater public understanding of the subject.
This talk will explore the implications of hyper-realistic generative modeling to computer art. As deep learning-based generative models like StyleGAN2 and GPT-3 inflate to unprecedented sizes, they become capable of producing photorealistic imagery and Turing-test level text, opening up numerous and controversial applications.

We'll show how some of these techniques have been re-purposed for artistic exploration lately, as well as review some of their more creative possibilities.
Ian is a Chief Data Scientist and Coach, he's helped co-organise the annual PyDataLondon conference with 700+ attendees and the associated 11,000+ member monthly meetup.

He runs the established Mor Consulting Data Science consultancy in London, gives conference talks internationally often as keynote speaker and is the author of the bestselling O'Reilly book High Performance Python (2nd edition).

He has 18 years of experience as a senior data science leader, trainer and team coach. For fun he's walked by his high-energy Springer Spaniel, surfs the Cornish coast and drinks fine coffee. Past talks and articles can be found at:
"My Pandas is slow!" - I hear that a lot. We'll look at ways of making Pandas calculate faster, help you express your problem to fit Pandas more efficiently and look at process changes that'll make you waste less time debugging your Pandas.

By attending this talk you'll get answers faster and more reliably with Pandas.
Vitaliy is from Kharkiv, Ukraine
He works in IT for more than 15 years. 12+ with Python
Vitaliy participated in various types of projects with high performance requirements.
Graduate of Belarusian State University (Faculty of Radiophysics and Computer Technologies).

Developer, Tech Lead, Researcher, Trainer at ISsoft and Coherent Solutions. Machine Learning group and Intelligence Solutions Department co-founder.
Aspect-based Sentiment Analysis is a challenging task of text analysis in e-Commerce. It researches and develops related ideas for text Sentiment Analysis to obtain Sentiment Analysis for each aspect in a text.

The proposed approach provides algorithms for aspects spotting based on semantic analysis of a text and estimating sentiment for the context of the aspects. Approach capabilities are perfectly matched for goods review analysis. In the presentation, it will be considered the solution based on BERT transformation and the development of a service for deploying the resulting model.
Python core developer, Python 3.8 and 3.9 release manager, creator of Black, pianist, dad.

Likes building synthesizers, immersive single-player role playing games like Fallout, and single malt Scotch whisky.
Programming languages with dynamic type systems as well as freeform document databases give us a big productivity boost early on in the project.

On the other hand, strong static type systems and relational databases protect us against an entire category of programming mistakes, and aid in code maintenance by helping IDEs auto-complete, jump to definition, refactor, and so on.

Is there a workable compromise? Can we somehow reap the benefits of both? Let's find out!
MinJae (a.k.a mingrammer) is the creator of Diagrams.
He has used Python for a long time to build automation and productivity tools.
Diagrams let you draw the cloud system architecture in Python code. It was born for prototyping a new system architecture design without any design tools.

In this talk, I will introduce the Diagrams as a whole and share why I made the Diagrams, and how it was designed and built.
Thomas is the CEO at PyMC Labs ( Prior to that Thomas was the VP of Data Science at Quantopian, where he used probabilistic programming and machine learning to help build the world's first crowdsourced hedge fund.

He is an author of the popular PyMC3 package — a probabilistic programming framework written in Python. He holds a PhD from Brown University.
Bayesian modeling is an extremely powerful tool in solving data science problems across different domains. And while user-friendly modeling packages like PyMC3 exist, understanding the underlying concepts still provides a challenge for many newcomers. The main reason is that usually, statistics is taught by statisticians who provide formulas with little regard for intuition.

In this talk I will take the opposite approach: throw all math out the window and explain the underlying concepts in an intuitive way.
Emeli is a Co-founder and CTO at Evidently AI, a startup developing tools to analyze and monitor the performance of machine learning models.
Earlier, she co-founded an industrial AI startup and served as the Chief Data Scientist at Yandex Data Factory. She led over 50 applied ML projects for various industries - from banking to manufacturing. Emeli is a data science lecturer at GSOM SpBU and Harbour.Space University.

She is a co-author of the Machine Learning and Data Analysis curriculum at Coursera with over 100,000 students.

She also co-founded Data Mining in Action, the largest open data science course in Russia.
ML-based services are different from traditional software applications. Once they are in production, we have to keep an eye not only on the code, but also on the data and model performance.

In this talk, I will share what new failure modes to prepare for, and how to set up model monitoring using open-source tools.
Artem develops projects on ML implementation in operational audit processes to optimize them and improve their user-friendliness. The key projects at Sberbank are the ones aimed to develop personalized recommendations for employees (from training to recommendations not to make mistakes).

Passionate about recommendations, applies them everywhere.
Are you tired of counting ratings in your recommendation systems yet? I am, so I started looking for other options in building recommendation systems.

We need to take a step away from "dry" ratings and try to understand the surrounding contexts. This type of recommendation changes the rating calculation from the user * item> rating formula to the user * item * context> rating formula, it seems very simple, but it is not!

How to do this and where to start? Let's work together along the path of adding context to recommendations. In my practice, adding context has helped increase conversions, but what will you get?
Alejandro is the Chief Scientist at the Institute for Ethical AI & Machine Learning, where he leads the development of industry standards on machine learning explainability, adversarial robustness and differential privacy.

Alejandro is also the Director of Machine Learning Engineering at Seldon Technologies, where he leads large scale projects implementing open source and enterprise infrastructure for Machine Learning Orchestration and Explainability.

With over 10 years of software development experience, Alejandro has held technical leadership positions across hyper-growth scale-ups and has delivered multi-national projects with top tier investment banks, magic circle law-firms and global insurance companies. He has a strong track record building cross-functional departments of software engineers from scratch, and leading the delivery of large-scale machine learning systems across the financial, insurance, legal, transport, manufacturing and construction sectors (in Europe, US and Latin America).
The lifecycle of a machine learning model only begins once it's in production. In this talk we provide a practical deep dive on best practices, principles, patterns and techniques around production monitoring of machine learning models. We will cover standard microservice monitoring techniques applied into deployed machine learning models, as well as more advanced paradigms to monitor machine learning models through concept drift, outlier detector and explainability.

We'll dive into a hands-on example, where we will train an image classification machine learning model from scratch, deploy it as a microservice in Kubernetes, and introduce advanced monitoring components as architectural patterns with hands-on examples. These monitoring techniques will include AI Explainers, Outlier Detectors, Concept Drift detectors and Adversarial Detectors.

We will also be understanding high level architectural patterns that abstract these complex and advanced monitoring techniques into infrastructural components that will enable for scale, introducing the standardised interfaces required for us to enable monitoring across hundreds or thousands of heterogeneous machine learning models.
Moshe has been using Python since 1998. They are a founding member of the Python Software Foundation and a founding member of the Twisted project.

They are interested in site reliability engineering and DevOps practices.
The strengths and weaknesses of Python lend themselves to a different style of object-oriented programming.

By accepting several constraints on how we design and implement classes, we make our code more robust, more testable, and easier to adapt to changing circumstances
I became a programmer, driven by a childhood dream of making video games.

During a long journey through software development, I made my bones in the wasteland of system, web and desktop applications. My dream came true three years ago, when I joined Wargaming and became part of the awesome team behind the legendary World of Tanks game. I am currently developing the gameplay logic for it.
Mid-80s. The game industry was relaunched with the release of Nintendo Entertainment System (aka "Dendy"), the 8-bit console which defined gaming for the decade to follow.
For its time, it was quite a piece of hardware, featuring the legendary 6502 CPU, 2KiB RAM, and a "GPU" capable of rendering a bunch of sprites on a 240x256 pixels TV screen.
How you ever make a game with that?! Back in time, it was raw assembly and thinking by hex numbers.

In this talk, I'll show you how the NES worked and how games for it were made.But to keep your sanity in place, there'll be almost no assembly. Instead, you will learn how to make your own compiled programming language in Python, to make the gamedev fun again!

I'll show how to define the grammar, parse it, translate the high-level constructs into NES-compatible machine code and build the final executable which goes into the ROM.
Press "start" to continue!
Andy Fundinger is a senior engineer at Bloomberg, where he develops Python applications in the Data Services Platform Scale and Reliability team. He also supports Python users throughout the firm via the company's Python Guild. Andy has spoken multiple times at Python and other developer conferences, including EuroPython, QCon, PyCaribbean, PyCascades, PyCon, PyGotham, and PyLondinium.

In the past, Andy has worked on private equity and credit risk applications, web services, and virtual worlds.

Andy holds a Master's in Engineering from Stevens Institute of Technology. He is the father of a 4-year-old
Many of the systems we build fail to account for performance, error rates, data correctness, or other customer reliability needs.

This talk discusses the basics of service level objectives by comparing them to ordinary functional specifications. It also serves as an introduction to System Reliability Engineering (SRE).
Andrei is a principal machine learning engineer and data scientist with primary expertise in natural language processing and information retrieval. In R&D department of IHS Markit in Minsk he works as the key developer of core machine learning components for new products and as the tech lead who drives engineering culture and innovative approaches.

One of the key contributor to the project that made his team one of the few first in the world to adopt SOTA transformer-based models in production information retrieval system. Andrei is formerly head of machine learning and advanced analytics sector in one of the largest IT services provider and has a wide experience in different fields of software engineering.
There are a lot of myths and misconceptions around applying best practices in the field of machine learning and data science. Also there is a big difference between one team of two DS who uses Git plus DVC and several teams of developers with diverse backgrounds and expertise.

Why and how to adopt best practices for machine learning projects. How to automate continuous integration and ML-models delivery, to get better codebase structure and to do data management. Why to do code review of experimental code. In this talk I will share our experience.
The speakers will discuss what they have discovered in the Python ecosystem recently and how it has improved their lives, as well as what announcements they are looking forward to and enthusiastic about.

What about you? What are you already using in production, and what else are you looking forward to on the next project? Join our chat, share your discoveries and achievements in Python.

The session is in Russian.
Each IT team accumulates its own life hacks and tricks, finds its favorite tools and processes. The speakers will talk about how their teams are solving typical problems faced by developers.

How do you do code review? How do you decide if you should start using FastAPI in your next project, or try Django for now? How do you document your code? We will ask our experts what they are doing and share our stories and findings in the chat.
I have more than 12 years of Python experience behind me, most of which was devoted to Python and preaching good coding practices.

From monolithic accounting systems to authentication microservices, it was a rough ride littered with rotting husks of legacy codebases. This path allowed me to accumulate bumps and some good advice which I shared previously on conferences and meetups. But there is always more to tell and to share.
Roman is CTO at Flo Health, #1 mobile product for women's health сhosen by over 165M users worldwide.

He is the leader of a team of more than 100 senior engineers with an average of 10 years of experience working with data-intensive applications, high-loaded systems, and complex infrastructures. Roman's team leverages machine learning and AI to provide accurate predictions and relevant, personalized, medically credible health tips for users. His team manages sensitive data to design, develop, and improve security services and data engineering processes.

Roman used to be a Head of Engineering at Payfort (Amazon service), he supervised the infrastructure, design, and architecture of a modern API-first payment gateway that was later acquired by Amazon. Roman also was in charge of tech at Adform Data Management Platform (DMP).

Roman is a respected member of IT/startup sector and participates in public events as a speaker and organizer. He leads several local developer communities in Belarus.

Follow us for the updates!

Sigining up the newsletter indicates you agree with our Privacy Policy. If you decide that you no longer want to receive a newsletter, you can unsubscribe by clicking the "Unsubscribe" link, located at the bottom of each newsletter.

Programme committee

Passionate Software Engineer
Software Engineer
Director of Engineering
Python developer who loves web
Principal Machine Learning Engineer
Data Scientist
Software Engineer

General Partner
ISsoft is an international IT company founded in 2004 as a subsidiary of Coherent Solutions Inc, recognized by Software 500, Fast 50, Inc. 5000, Clutch Top-100, official partner with Microsoft, Xamarin and Amazon Web Services.

Company strives to create a work environment that is both productive and comfortable. What brought them to the top are ISsoft's people. They value every aspect of their contributions and strive to make ISsoft a place for career growth and opportunities.

ISsoft focuses on top-notch IT-solutions for North American and European markets. 100 of clients trust ISsoft to work on product software development, IT-consulting, data and analytics, machine learning, mobile app development, DevOps and cloud, Salesforce development and more.

Writing code for pleasure. 14 years in software engineering, 7 of them — with Python.

Organizer of the Minsk Python Meetup and sub-events around.
Python developer at day time, Go developer (gopher) under the hood. Big fan of full-text search and graph databases.

Contributed to different python/go open source projects:
- pyhelm, aiohttp-swagger, mezzanine
- chalice, requests, aiohttp tutorial
- sendgrid-python and sendgrid-django
- OpenAPI v3 specification, fix Go docs

Speaker at PyCaribbean, PyCon Italia 2017, EuroPython 2016, PyCon Ukraine 2014, PyCon Belarus 2015-2018 PyCon Russia 2015, 2016.
Blogger at

Passionate software engineer. Digital nomad. Python developer.
Area of interests: high load applications.
Python and Rust enthusiast from Minsk, Belarus.

Passionate about communities, artificial intelligence and development.

Now building the world's most advanced Enterprise AI platform at DataRobot.
Roman writes Python code for more than a decade. Occasionally he shares his experience in blog posts and talks.

An active member of Python communities in Russia, Belarus, and Portugal.
Organized PyCon conferences in Russia and Belarus and various Python events in Porto.

Roman believes that programming is a skill that opens many doors and is eager to help people follow their passion for becoming developers.
A seasoned machine learning engineer with 15 years of industry experience in data science and software architecture.

His primary interests are productionalizing data science, AutoML, time series forecasting, and processing spoken and written language.

He teaches AI and ML at UCU, competes on Kaggle, and has led multiple international data science and engineering teams.

Yuriy currently works as Principal Machine Learning Engineer at DataRobot.
Data Science and ML enthusiast with more than 12 years of industry experience in different capacities and with vast expertise in insurance, e-commerce, marketing, sports and e-sports domains.

Right now works in DataRobot and leads team responsible for delivering high complexity data science use cases for customers in USA and Europe

With the support of

The mission of the Python Software Foundation is to promote, protect, and advance the Python programming language, and to support and facilitate the growth of a diverse and international community of Python programmers. The majority of the PSF's work is focused on empowering and supporting people within the Python community. The PSF has active grant programs that support sprints, conferences, meetups, user groups, and Python development efforts all over the world. In addition, the PSF underwrites and runs PyCon US, the primary Python community conference.

Being part of the PSF means being part of the Python community. Recently the PSF has been changed to an open membership organization, so that everyone who uses and supports Python can join.

Special Partner

Wargaming (JLLC Game Stream – the development studio in Minsk) is one of the biggest online game developer and publisher headquartered in Nicosia, Cyprus. We are not only about "Tanks". We created "World of Warships", "World of Tanks Blitz" and more than 10 other games with the audience of 200+ million players on all major platforms. Our games were recognized with variety of awards, including four record-breaking Golden Joystick Awards. Operating since 1998, Wargaming has grown to become one of the leaders in the gaming industry with 5000+ employees and offices spread all over the world. Our flagship product – the massively popular free-to-play hit World of Tanks – was created in Minsk which still remains "the development headquarters" for the game.

Goods Partner

General Media Partner

Media Partners

We use cookies to provide the best website experience
OK, don't show again
PyConBY 2021 Conference Code of Conduct
All attendees, speakers, sponsors and volunteers at our conference are required to agree with the following code of conduct (CoC). Organisers will enforce this code throughout the event. We are expecting cooperation from all participants to help ensuring a safe environment for everybody.

PyConBY 2021 is a community conference intended for networking and experience exchange in the developers community.

PyConBY 2021 Conference is dedicated to providing a harassment-free conference experience for everyone, regardless of gender, sexual orientation, disability, physical appearance, body size, race, or religion. We do not tolerate harassment, discrimination, abasement and any form of disrespect.Sexual language and imagery is not appropriate for any conference venue, including talks.

We urge to avoid offensive communication related to gender, sexual orientation, disability, physical appearance, body size, race, religion, sexual images in public spaces, deliberate intimidation, stalking, following, harassing photography or recording, sustained disruption of talks or other events, inappropriate physical contact. Attending the event under the influence of alcohol or other narcotic substances is unacceptable.

Exhibitors in the expo hall, sponsor or vendor booths, or similar activities are also subject to the anti-harassment policy. In particular, exhibitors should not use sexualized images, activities, or other material. Booth staff (including volunteers) should not use sexualized clothing/uniforms/costumes, or otherwise create a sexualized environment.

Conference participants violating these rules may be sanctioned or expelled from the conference without a refund at the discretion of the conference organizers.

Expected Behavior
  • Participate in an authentic and active way.
  • Exercise consideration and respect in your speech and actions.
  • Attempt collaboration before conflict.
  • Refrain from demeaning, discriminatory, or harassing behavior and speech.
  • Be mindful of your surroundings and of your fellow participants. Alert organisers if you notice a dangerous situation, someone in distress, or violations of this CoC.
Thank you for helping make this a welcoming, friendly event for all!

Need Help?
Contact the organizer at