All the talks and speakers at a glance

See also the Detailed Schedule.

Core Python (Day 1 & 2)

Web & Cloud (Day 1 & 2)

Data 1 - Integrating and shaping data (Day 1 only)

Data 2 - Learning, deep and wide (Day 1 only)

scikit-learn day (Day 2 only)

Education (Day 2 only)

Tutorials / Workshops (Day 2 only)

Details of each talk

Alexander Hendorf: Neat Analytics with Pandas Indexes

Abstract:

Pandas is the Swiss-Multipurpose Knife for Data Analysis in Python. In this talk we will look deeper into how to gain productivity utilising Pandas powerful indexing and make advanced analytics a piece of cake.

Speaker: Alexander Hendorf (Königsweg GmbH)

Speaker bio:

Alexander’ professional carrer was always about digitalisation: starting from vinyl recors in the nineties to to advanced data analytics nowadays. He’s program chair and organiser of Europe’s main Python conference EuroPython and 2017 PyConDE, one of the 25 mongoDB masters and a regular contributor to the tech community. He has spoken at many international conferences in Silicon Valley, New York, London, Florence or Kiew. He’s working at Königsweg a consultancy for start-ups and the industry. Königsweg enhances your entrepreneurial impact through the development and implementation of high-performing transformation strategies across all business divisions.

Alexandre Abadie: Cloud computing made easy in Joblib

Abstract:

Joblib is a Python package initially designed for efficient computing of embarrassingly parallel problems on a local computer or a laptop. This talk gives a short introduction of the features provided by Joblib and the recent developments that make them usable on Cloud computing infrastructures.

Speaker: Alexandre Abadie (Inria)

Speaker bio:

Alexandre Abadie: Using Python for IoT: a return of experience

Abstract:

This is a return of experience of how we built an Open Source application thanks to Python asyncio for interacting in real-time with low-end constrained IoT nodes. The application only uses standard IoT protocols such as CoAP or MQTT and nodes resources are displayed on a dynamic web dashboard.

Speaker: Alexandre Abadie (Inria)

Speaker bio:

Anaël Bonneton: Machine Learning for Computer Security Experts using Python & scikit-learn

Abstract:

We present SecuML, a Python open source tool that aims to foster the use of Machine Learning in Computer Security. It allows security experts to train models easily and comes up with a user interface to visualize the results and interact with the models.

Speaker: Anaël Bonneton (ANSSI - ENS Paris)

Speaker bio:

I am a PhD student jointly supervised by Francis Bach (INRIA, SIERRA team), and Pierre Chifflier (ANSSI, French Network and Information Security Agency). I work on Machine Learning applications to Computer Security and I focus on interactive Machine Learning (active learning and rare category detection) to allow security experts interact with models. I am a strong supporter of open source solutions to facilitate comparisons in research works, but also to strengthen the use of Machine Learning by application domain practitioners who are not Machine Learning experts.

Antoine Pietri: Camisole : a secure online sandbox to grade students

Abstract:

Camisole is an online sandbox: you give some untrusted source code, it compiles it and runs it against your tests. Learn how to build a secure sandbox with Linux kernel features, and how you can use it to teach programming, grade your students and organize programming competitions!

Speaker: Antoine Pietri (Prologin)

Speaker bio:

I’m a software engineer working at Software Heritage (Inria project) as a Python developer. I have been working with Python for as long as I remember, and I love to use all the shiny new features of the language. I have been organizing Prologin, the French national programming contest, for more than 6 years and gained a lot of experience there working with untrusted code execution, infrastructure development and network matchmaking frameworks, and writing a lot of algorithmic problems for the competition.

Benjamin Guinebertière: Program in Python against big data clusters from one VM, thanks to Docker

Abstract:

You want to develop in Python against big data clusters but you don’t want to run a whole infrastructure? During this workshop, we’ll run boontadata samples starting from a blank Virtual Machine, with clusters like Apache Kafka, Cassandra, Spark or Flink for instance.

Speaker: Benjamin Guinebertière (Microsoft)

Speaker bio:

Benjamin Guinebertière works with startups and companies of different sizes to help them technically adopt Microsoft Azure cloud, should they use Big Data, Machine Learning or other technologies. He also speaks at conferences, writes (blogs, …) and takes feedback. In July 2016, as Microsoft embraces more and more open source, he started a project called boontadata that compares big data streaming engines from a functional and coding point of view. You can find Benjamin at 3-4.fr.

Benjamin Talmard: Serverless architecture in Python with Azure Functions

Abstract:

Come to learn more on how to leverage serverless architecture in your Python based solutions. From IoT to APIs including async taks in your web applications, serverless can change your mind about the Cloud.

Speaker: Benjamin Talmard (Microsoft France)

Speaker bio:

As the CTO in residence of Microsoft Accelerator Paris, I’m working with tens of startups building solution on the Cloud in Python. My job is to help them built the right software architecture and use the right services to help them meet their scalability, availability and performance requirements.

Bhargav Srinivasa Desikan: Topic Modelling (and a lot more) with NLP framework Gensim

Abstract:

This tutorial is going to guide you through the process of analysing your textual data through topic modelling - from finding your data, pre-processing, applying topic modelling algorithms - before moving on to more advanced textual analysis techniques. Text mining has never been so easy.

Speaker: Bhargav Srinivasa Desikan (INRIA, France)

Speaker bio:

I’m a student researcher currently working at INRIA, France. I’m part of the MODAL (Models Of Data Analysis & Learning) team, and I work on Metric Learning, Predictor Aggregation and Data Visualisation. When I’m not at my lab (and sometimes when I’m at my lab), I enjoy contributing to open source - particularly the Python Machine Learning and Scientific computing community. I previously participated in Google Summer of Code 2016, where I implemented Dynamic Topic Models for Gensim. I continue contributing regularly to a variety of python machine learning libraries. I love attending and speaking at PyCons and PyDatas - I’ve previously spoken at PyCon France 2016, PyCon Slovakia 2017, and will be heading down to Vienna to speak at PyDays 2017. Come for the python, stay for the awesome community! When I’m not coding or working on ML research I enjoy drinking cold beer, reading science fiction and playing video games.

Boris FELD: Unicode and bytes demystified

Abstract:

Unicode and bytes were hard to work with in Python 2, luckily Python 3 came and put some order in this bazar. Alas the old habits die hard and migrating from Python 2 to Python 3 can often be hard without understanding all the concepts. Let’s review them for a happy Python 3 migration in 2017!

Speaker: Boris FELD ()

Speaker bio:

I’ve given several talks already which are available here: https://speakerdeck.com/lothiraldan, several of them were in English. I had to deal with unicode difference in Python 2 and Python 3 while packaging a wheel and so end up digging up in the PEPs and various RFC.

Burkhard Kloss: Performant Python

Abstract:

Python is a great language. Easy to learn, friendly, widely used. Not, however, fast. Usually this doesn’t matter. But when it matters, it really matters! How do we make our Python fast enough - but still keep it readable?

Speaker: Burkhard Kloss (Applied Numerical Research Labs)

Speaker bio:

I only came to England to walk the Pennine Way… 25 years later I still haven’t done it. I did, though, get round to starting an AI company (spectacularly unsuccessful), joining another startup long before it was cool, learning C++, and spending a lot of time on trading floors building systems for complex derivatives. Sometimes hands on, sometimes managing people. Somewhere along the way I realised you can do cool stuff quickly in Python, and I’ve never lost my fascination with making machines smarter.

Fabrice Salvaire: Circuit Simulation using Python

Abstract:

PySpice is a module which make the bridge between the Berkeley SPICE electronic circuit simulator and the powerful scientific ecosystem of Python, so as to steer the simulation and analyse its output. This project targets especially an education or DIY audience and provides educational contents.

Speaker: Fabrice Salvaire ()

Speaker bio:

I am a project manager in the industry and I designed scientific softwares using Python and its ecosystem. I participated to the PyConFr 2014 conference where I given a talk on a newer OpenGL binding design and I written an article in the proceedings on Euroscipy 2013: High-Content Digital Microscopy with Python, Fabrice Salvaire, Part of the Proceedings of the 6th European Conference on Python in Science (EuroSciPy 2013), Pierre de Buyl and Nelle Varoquaux editors, (2014) arXiv:1404.6385v2. As an open source evangelist, I written many softwares using Python which are all available on my Github page, including these ones: PyOpenGLng: an experimental wrapper for the OpenGL API, pyglfw-cffi: A wrapper for GLFW based on CFFI, PyDvi: a module to read and process DVI (DeVice Independent) files, the native output of TeX, CodeReview: a Git GUI tool to perform code review using the Qt5 framework, PyValentina: a Python implementation of the Valentina Pattern Drafting software, A Xournal clone with GPU acceleration, A web site based on Django and Flask, some DjangoCMS plugins.

Francis Wolinski: Introduction to Data Analysis using Python

Abstract:

The aim of the workshop is to put into action the Pandas Python library to deal with real data. It is intended for beginners and will be based on a notebook with datasets to perform in-depth analysis. Basic to advanced functionalities of Pandas and Matplotlib will be experimented during the session.

Speaker: Francis Wolinski (Yotta Conseil)

Speaker bio:

My name is Francis Wolinski. I have been a computer scientist for more than 25 years. I have been using Python for more than 12 years having acquired vast experience in various fields: banking and insurance, auditing and accounting, bioinformatics and natural language processing. Currently, I am managing consulting activities in Information Technology and Data Science. I am also teaching and training the Python Ecosystem for Data Science in prestigious institutions such as the “Université Paris 1 Panthéon-Sorbonne“ and the “Banque de France“.

François Sausset: From Python to smartphones: neural nets @ Saint-Gobain

Abstract:

We use Keras for fast prototyping neural nets in a variety of applications. To put them in the hands of end-users, these models are ported to smartphones GPUs using Tensorflow. As an example, we built a generic prototype app able to handle any input tensors, in particular for audio recognition.

Speaker: François Sausset (Saint-Gobain Recherche)

Speaker bio:

Data science team leader @ Saint-Gobain. PhD in statistical/theoretical physics.

Gael Varoquaux: Scikit-learn: an incomplete yearly review

Abstract:

I will give an update on the scikit-learn project: new and ongoing features, code improvements, and ecosystem.

Speaker: Gael Varoquaux (Inria)

Speaker bio:

Gaël Le Mignot: Ways to generate PDF from Python Web applications

Abstract:

An overview of different tools available to generate and manipulate PDF from Python Web applications (weasyprint, reportlab, LaTeX, pdftk, …) with real-life examples, accompanied with tips and pitfalls to avoid.

Speaker: Gaël Le Mignot (Pilot Systems)

Speaker bio:

Gaël Le Mignot is a Python developer, Debian sysadmin and project manager at Pilot Systems, working since more than 12 years in Python Web development, mostly with the Zope and Django frameworks. Previous talks include various editions of Plone Conference, Djangocong, RMLL/LSM, and Pgdays Paris, as well as computing science schools (EPITA, INSA).

Igor Mosyagin: Machine Learning in computational materials science: an overview, a primer, and a rant.

Abstract:

An overview of the current trends in modern computational materials science and the cross-pollination with DataScience community. I’m going to present a few case studies and give a brief introduction of my own experience with using python in academic community.

Speaker: Igor Mosyagin (Materials Modeling laboratory, NUST “Moscow Inistute of steels and alloys”)

Speaker bio:

Pragmatic programmer with strong scientific background. I have two PhDs in theoretical physics and do web dev in python for fun. I try to force my academic colleagues to be a better humans by being better programmers.

Jan Margeta: The rod of Asclepios: Machine learning in Python for cardiac image analysis

Abstract:

Python is a superpower solving the hardest data challenges. Let me share with you our experiences in building machine learning system prototypes for medical image analysis and deploying them to production. Join us on our journey at KardioMe and see how simple it is to start analysing your data too.

Speaker: Jan Margeta (KardioMe)

Speaker bio:

Jan is the founder of KardioMe, a Python aficionado, and a white water kayaker. He did his PhD in machine learning for automated medical image analysis at Inria Sophia Antipolis, Microsoft Research Cambridge and MINES ParisTech and a master in computer vision and robotics. Now, he is putting all the research experience into real-world use to improve how we treat cardiac diseases, to make our healthcare a bit more efficient, and to engage patients with their own health. Jan is passionate about using technology to push the boundaries of human knowledge, teaching computers to see, solving hard challenges with data, and making our planet a sustainable place.

Jarosław Szymczak: Automatic image moderation in classifieds

Abstract:

In image moderation, the unwanted images do not have to be commonly referred to as illicit. In classifieds industry what is accepted and what not is very relative. We created a solution that handles image and contextual data together, to improve the content quality of OLX sites across the globe.

Speaker: Jarosław Szymczak (OLX Group (Naspers Services GmbH))

Speaker bio:

Machine Learning Scientist in OLX Tech Hub Berlin. Having background in analytics and predictive models creation for finance institutions, FMCG and Telecom companies. Currently specializing in applying machine learning to detection of unwanted content on OLX classifieds sites across the globe.

Jean-Baptiste Aviat: Writing a C Python extension in 2017

Abstract:

This talk describes the build of a C Python extension, with prebuilt binaries, when modern packaging standards, as well as Docker, have been a game changer in the Python extensions world. Many examples come from our experience building PyMiniRacer, used in production across hundreds of companies.

Speaker: Jean-Baptiste Aviat (Sqreen)

Speaker bio:

Jean-Baptiste Aviat is CTO at Sqreen. He spent half a decade hunting vulnerabilities at Apple, helping developers solve them, and developing security software.

Jean-François Bercher: Some extensions for Jupyter/IPython notebook

Abstract:

This talk will present and describe some Jupyter/IPython notebook extensions by the author and others of the jupyter_contrib_nbextensions group, which are useful for - core development and data analysis, - publication of research reports and papers, - classroom animation.

Speaker: Jean-François Bercher (ESIEE Paris)

Speaker bio:

Jean-François Bercher is Professor at ESIEE Paris, where he teaches (statistical) signal processing, machine learning, programming, … He is the head of the engineer track in DataScience, networks and IoT. He is also the the Dean of the Faculty. As for Research, he works in the field of information theory, around information mesures, divergences and generalized entropies. Advocating the use of Python both in research and teaching, he has developed several extensions to the Jupyter/IPython notebook and is one of the mainteners of jupyter_contrib_nbextensions repository.

Jean-Paul Smets: Improving python memory management in a multi-process environment with wendelin.core

Abstract:

This talk explains how to efficiently use memory with mutli-process deployments of cpython on multi-core servers or clusters, thanks to a distributed transactional memory manager: wendelin.core.

Speaker: Jean-Paul Smets (Nexedi)

Speaker bio:

Founder of Nexedi and initial author of ERP5, Jean-Paul is currently in charge of Nexedi international development, investment relations and technology partnerships. Jean-Paul graduated from Ecole Normale Supérieure with a PhD in computer science and from Ecole des Mines de Paris with a Master in Public Administration. He gained industrial experience in Apparel Industry, Oil industry, Non Profit Organizations and Lorraine Region Public Administration. He is an active member of Free Software associations and has played in key role in the Eurolinux campaign which succeeded in protecting innovation from software patents.

Jean-Sébastien Bevilacqua: Call a C API from Python becomes more enjoyable with CFFI

Abstract:

The objective of this call is to share the best way to write a C extension for Python. Lot of good libraries are written in C, a Python developer should know how to quickly write a C library wrapper. After this call, Python developers will fully understand taking and ending of Python extensions.

Speaker: Jean-Sébastien Bevilacqua (Linagora)

Speaker bio:

Jean-Sébastien Bevilacqua is 27 years old. He has a degree in computer engineering. He worked three years at Thales on OpenStack, where he has discovered Python and started loving it. He then joined Linagora to be engaged in the open source. He likes to talk with developers on IRC or github. Lastly, he is a contributor to libGDX and Vulkan ecosystem. He likes 3D and low-level development. He likes traveling and his favorite member of One Direction is Liam Payne ;-)

Jill-Jênn Vie: Introduce AI to kids? It’s possible!

Abstract:

Sentence generation, programming bots for simple games, digit classification: we will present 3 years of AI-related projects with kids, what worked and what did not. Can you convey what you’re working on to a nice challenge for a kid?

Speaker: Jill-Jênn Vie (RIKEN)

Speaker bio:

Jill-Jênn Vie has learned programming through contests, then programmed contests for learning. (But now he contests that he actually learned programming…) He lives in Tokyo and does ML research on educational data. Know more!

Johan Mabille: xtensor: the lazy tensor algebra library

Abstract:

xtensor is a C++ template tensor algebra library supporting numpy-style broadcasting and universal functions. In this talk we present the highlights of the expression system, then we show how xtensor can be used to create numpy-aware Python extension modules with the xtensor-python project.

Speaker: Johan Mabille (QuantStack)

Speaker bio:

Johan Mabille is a scientific software developer specializing in high-performance computing in C++. He holds master’s degree in computer science from Centrale-Supelec. As an open source developer, Johan coauthored xtensor and xeus , and is the main author of xsimd. Prior to joining QuantStack, Johan was a quant developer at HSBC.

Joir-dan Gumbs: Syncing up with Python’s asyncio for (micro) service development

Abstract:

Thinking about asyncio for your next project? Looking to leave flask for a younger, more asynchronous framework? Then join me as we explore features of asyncio with examples centered around building asynchronous (micro) services!

Speaker: Joir-dan Gumbs (IBM)

Speaker bio:

I am a traveled Senior Software Engineer at IBM. Currently residing in San Francisco, my primary passion is games; specifically around game analytics, and backend services that power games. This includes interests in container platforms, event-driven micro services, ETL pipelines, and game report generation. Although a polyglot (in both programming and natural languages), I prefer Python for flexibility, power, and educational ease. Languages: English (native), Japanese (I get by ^_^), Spanish (beginner), French (beginner) Video Games: Overwatch (PS4), DDR Ace, SSB4 Sports: Basketball, Football (U.S.), Badminton

Joris Van den Bossche: Data analysis with Pandas

Abstract:

In this hands-on tutorial on using pandas for data analysis, you will be guided through some of the powerful methods and concepts in pandas, including time series manipulation (resampling and rolling operations), groupby operations, reshaping with stack/unstack/pivot, …
This tutorial will go deeper in a selection of concepts and it assumes already basic familiarity with pandas (or e.g. having followed the beginners tutorial in the morning).

Speaker: Joris Van den Bossche (Université Paris-Saclay Center for Data Science, INRIA)

Speaker bio:

I did a PhD at Ghent University and VITO in air quality research. In addition, I am also a core developer of Pandas, the main data analysis library in Python. Currently, I am working at the Paris-Saclay Center or Data Science.

Joris Van den Bossche: Pandas: what’s new and what’s coming

Abstract:

Overview of what’s happening in the latest releases of pandas and where we are heading to.

Speaker: Joris Van den Bossche (Université Paris-Saclay Center for Data Science, INRIA)

Speaker bio:

I did a PhD at Ghent University and VITO in air quality research. In addition, I am also a core developer of Pandas, the main data analysis library in Python. Currently, I am working at the Paris-Saclay Center or Data Science.

Josef Spillner: Function-as-a-Service: A Pythonic Perspective on Serverless Computing

Abstract:

With Function-as-a-Service (FaaS), individual functions process discrete data volumes in the cloud. FaaS services are offered commercially with a pay-per-call pricing. This tutorial introduces alternative open FaaS frameworks and shows how to use them for writing functions in Python 2 and 3.

Speaker: Josef Spillner (Zurich University of Applied Sciences, Switzerland)

Speaker bio:

Josef Spillner is a senior lecturer and head of the Service Prototyping Lab at Zurich University of Applied Sciences in Switzerland. Teaching basic Python to 220 students a year of which 80 advance into more complex Python application projects in the subsequent semester, he also sees the language and its ecosystem widely used in the lab for prototyping software as measurable and demonstratable research output. Before founding the lab, he conducted research and led activities as post-doc at TUD, SAP, NTUU, UFCG and UniBZ, wrote a doctoral dissertation about metaquality of services and a habilitation treatise about stealth computing in multi-cloud environments. The latter work resulted in StealthDB, another prototype implemented in Python 3.

Julien SIMON: An introduction to Deep Learning with mxnet

Abstract:

Using Python notebooks, we’ll show how to start writing Deep Learning applications using Mxnet, a popular library for Deep Learning which support both CPUs and GPUs. Then, we’ll demonstrate how you can achieve excellent results quickly by using state-of-the-art pre-trained models.

Speaker: Julien SIMON (Amazon Web Services)

Speaker bio:

Julien Simon, Principal Technical Evangelist at Amazon Web Services. Before joining AWS, Julien served for 10 years as CTO/VP Engineering in top-tier web startups. Thus, he’s particularly interested in all things architecture, deployment, performance, scalability and data. As a Principal Technical Evangelist, Julien speaks very frequently at conferences and technical workshops, where he meets developers and enterprises to help them bring their ideas to life thanks to the Amazon Web Services infrastructure.

Loryfel Nunez: How to prepare your text data for NLP applications

Abstract:

Data cleaning is not as sexy as the the actual NLP algorithms. True. BUT, how your prepare your data will determine how well, or poorly your algorithm will perform. This talk will focus on Python’s libraries to extract important text or to remove unwanted text to prepare your data for NLP tasks.

Speaker: Loryfel Nunez (FindSignal)

Speaker bio:

I have been in the intersection of research and engineering in the last 12 years. I am currently the Lead Data Engineer for a startup based in NYC. We are partnering with one of the world’s largest News organization to ingest and analyze their data. In my free time, I am a parent volunteer in my son’s high school. I am teaching myself how to knit and how to play the piano as I am preparing for the empty-nest stage in my life as my son now heads to college. I love hip hop and R&B.

Ludovic Gasc: AsyncIO and aiohttp workshop

Abstract:

You want to learn or discover AsyncIO in general and play with aiohttp as example with WebSockets ? You are in the right workshop :-)

Speaker: Ludovic Gasc (ALLOcloud)

Speaker bio:

Ludovic Gasc is the lead architect of ALLOcloud, a highly renowned open source VoIP and unified communications company in Europe. Over 7 years, Ludovic has developed redundant distributed systems for Telecom, based on Python, AsyncIO, WebRTC and PostgreSQL.

Lynn Cherny: Exploring French Job Ads

Abstract:

To address some faculty questions at EM-Lyon Business School, some students and I began collecting and analysing job ads for French companies. I’ll show some preliminary text analysis and visualization of the content, focused on skills sought by French employers (both business and technical).

Speaker: Lynn Cherny (em-lyon business school)

Speaker bio:

A former consultant in data analysis an visualization, Lynn Cherny is now teaching introduction to data science classes at EM-Lyon Business School. Her interests are interactive data vis, NLP, machine learning. She has a Ph.D. from Stanford in Linguistics that’s probably older than you.

Maciej Polańczyk: Mock it right! A beginner’s guide to world of tests and mocks.

Abstract:

The main goal of this presentation is to make beginner audience familiar with mocks and patches. I will say what are the best practices and what kind of mistakes, done by many developers, may be avoided so the dragons will not come. During this presentation I will prepare mock-it-o drink as a prize.

Speaker: Maciej Polańczyk ()

Speaker bio:

I started my programming journey with Basic and C++. I spent few years with Java and finally I discovered Python. I’m keen on having good quality tests. For few years I was working for TomTom in the team responsible for CI and this was a good lesson of how unit tests should be written to provide best information in case of failure. Currently, I work for STX Next where I am developing web services and sharing knowledge about best unit testing practices.

Margriet Groenendijk: A Beginners Guide to Weather & Climate Data

Abstract:

Weather is part of our every day lives. Who doesn’t check the weather forecast before heading out? But where does this data come from, what is it made of? This session looks at weather observations and models. Learn from examples how you can use weather and climate data yourself.

Speaker: Margriet Groenendijk (IBM Watson Data Platform)

Speaker bio:

Margriet is a Data Scientist and Developer Advocate for the IBM Watson Data Platform. She has a background in Climate Science where she explored large observational datasets of carbon uptake by forests and the output of global scale weather and climate models. Now she uses this knowledge to create clear plots and models from diverse data sets using cloud databases, data warehouses, Spark, and Python notebooks.

Maxime Beauchemin: Advanced Data Engineering Patterns with Apache Airflow

Abstract:

Analysis automation and analytic services is the future of data engineering! Apache Airflow’s DSL makes it natural to build complex DAGs of tasks dynamically, and we’ll explain how Airbnb has been leveraging this feature in intricate ways, creating a wide array of services as dynamic workflows.

Speaker: Maxime Beauchemin (Airbnb)

Speaker bio:

Maxime Beauchemin works at Airbnb as part of the “Analytics & Experimentation Products team”, developing open source products that reduce friction and help generating insight from data. He is the creator and a lead maintainer of Apache Airflow [incubating] (a workflow engine), Superset (a data visualization platform), and recognized as a thought leader in the data engineering field. Before Airbnb, Maxime worked at Facebook on computation frameworks powering engagement and growth analytics, on clickstream analytics at Yahoo!, and as a data warehouse architect at Ubisoft.

Moisés Guimarães de Medeiros: wolfcrypt: wrapping secrets in Python

Abstract:

The art of keeping secrets safe is one of the pillars of the internet. e-commerce, online banking, video conferences and a lot more are protected by encryption to become viable. In this presentation we will talk about modern cryptography concepts and use the wolfcrypt python module to apply them.

Speaker: Moisés Guimarães de Medeiros (wolfSSL)

Speaker bio:

Moisés Guimarães is a Brazilian pythonista who lives in João Pessoa, the capital of the state of Paraíba in Brazil, also the eastmost place of the Americas. Author of both wolfcrypt and wolfssl modules, CFFI wrappers for the wolfSSL crypto and SSL/TLS libraries. He works both as a software engineer at wolfSSL and as a professor at the Federal Institute of Education, Science and Technology of Paraíba, graduated in Web Development with a specialization degree in Information Security.

Nawfal Tachfine: Building and deploying a predictive API using scikit-learn, Flask and Docker.

Abstract:

The aim of this workshop is to expose a trained scikit-learn machine learning model as a REST API, built with Flask and Docker, to be queried by any system in JSON.

Speaker: Nawfal Tachfine (Aramisauto.com)

Speaker bio:

I come from a math-heavy background, with a focus on probability and statistics. I have also enjoyed writing code over the past 10 years. I am currently a data scientist at aramisauto.com, a French e-commerce company that buys and sells cars online, where I design and build end-to-end machine learning pipelines. Python has been my favorite language for years now. I am glad that I can use it on my day job.

Nicolas Hug: Collaborative filtering for recommendation systems in Python

Abstract:

In this talk we will present the topic of recommendation systems. We will focus on two popular approaches: neighborhood-based methods and matrix-factorization based techniques. We will also present Surprise, an open source Python package for easily building and analyzing recommendation algorithms.

Speaker: Nicolas Hug ()

Speaker bio:

I’m a PhD student in Machine Learning at IRIT lab in Toulouse, France. During my PhD I worked on recommender systems, which led me to develop Surprise, an open source Python library for building and analyzing recommender systems.

Phillip Schanely: Incremental Computation in Python

Abstract:

Our programs often end up running nearly identical workloads over only slightly changed inputs. We’ll identify this problem over many domains like UIs, databases, and build systems, and introduce a Python library for incrementally updating the results of a computation efficiently.

Speaker: Phillip Schanely (Google)

Speaker bio:

Phillip Schanely is your average programming languages nerd and works for Google in New York. His touch-based programming language made the frontpage of hackernews and reddit. Phillip will get very excited if you chat with him about software optimization, incremental computation, scoring functions, or socially acceptable tactics for writing code in bars.

Pierre Augier: FluidDyn, open-science in fluid dynamics with Python: examples of recent developments

Abstract:

Open-source and open-science in the field of fluid dynamics is an emerging trend which ensures transparency in reported results and better code development. The project FluidDyn fosters the development of Python packages in teaching and research with experiments, simulations and data processing.

Speaker: Pierre Augier (LEGI, CNRS, UGA, Grenoble INP)

Speaker bio:

CNRS Researcher in a laboratory in Grenoble (LEGI, Laboratoire des Écoulements Géophysiques et Industriels), I am specialized in turbulent flow regimes important for the atmospheric and the oceanic dynamics. I work with experiments in the Coriolis platform (a big rotating pool of 13 meter diameter), numerical simulations and data analysis of atmospheric models.

Pierre Poulain: Using Python to fight Malaria

Abstract:

Malaria is a mosquito-borne disease that killed 438,000 people in 2015. Using the Python ecosystem, we are building the first open database for blood smears used in malaria diagnosis, together with an E-learning plateform to train technicians. We also aim to develop automated diagnosis of malaria.

Speaker: Pierre Poulain (University Paris Diderot & CNRS)

Speaker bio:

I am a lecturer in computational biologist at the University Paris Diderot. I discovered Python during my PhD in Physics and never came back since. I had the chance to spend four year in the Republic of Congo, Central Africa. Over there I discovered malaria and the burden it represents for African populations. Back to Europe, I am trying to find smart technology- and data-based solutions to fight malaria.

Pierre-Alain Jachiet and Aurélien Gervasi: Open-Source Analytics Stack on MongoDB, with Schema

Abstract:

So your new application is gaining traction. It produces valuable data, stored in MongoDB. Time to exploit it ! But your BI tools and data scientists do not speak schema-less json… We expose here a full-fledged open-source stack to unlock the analytical potential of your MongoDB data.

Speaker: Pierre-Alain Jachiet and Aurélien Gervasi (OCTO Technology)

Speaker bio:

https://fr.linkedin.com/in/pierrealainjachiet https://fr.linkedin.com/in/aureliengervasi

Romain Dorgueil: Simple ETL in python 3.5+ with Bonobo

Abstract:

Simple is better than complex, right? True for data pipelines too. Bonobo is a python 3.5+ library used to code, run, monitor and maintain data pipelines. It’s plain, simple, modern, and atomic python. Spoiler : there is no «big data» here.

Speaker: Romain Dorgueil (L’Atelier)

Speaker bio:

Developer, sysadmin, technical team builder, founder of two companies), advisor. Currently helping start-ups to achieve more with less resources in our acceleration programs in Paris, and in charge of our product development activities. Sometimes, I play go and make music, but not at the same time.

Roman Yurchak: FreeDiscovery - information retrieval and e-Discovery in Python

Abstract:

This talk introduces FreeDiscovery - an open-source Python software that provides a REST API for information retrieval applications. Based on the scikit-learn machine learning library, its features include text categorization, semantic search, hierarchical clustering and duplicates detection.

Speaker: Roman Yurchak (Symerio)

Speaker bio:

Roman Yurchak is a data scientist at Symerio and lead developer of FreeDiscovery. Formerly he was a physicist working on predictive models in the field of laboratory astrophysics.

Sheer El Showk: Building a high-performance, scalable ML & NLP platform with Python

Abstract:

Python is the de-facto language for natural language processing (NLP) but scalability and performance bring their own unique challenges. In this talk I will outline the architecture and tools we used to build a production grade, scalable platform providing highly customizable NLP services.

Speaker: Sheer El Showk (Lore Ai)

Speaker bio:

As mentioned above I am currently the CTO and co-founder of Lore Ai, a US/France-based startup providing natural language products for business intelligence. Before founding Lore Ai, I was a theoretical physicist working on both string theory and statistical physics. I’ve held research positions at CEA Saclay, CERN and I now hold a permanent research position at UPMC, Paris 6 (from which I’m on leave). Part of my research required the development of novel linear optimization algorithms. I also implemented these in Python/Cython and built a parallelized version to run at scale on clusters with hundreds of cores (our computations often required dozens of core-years). The algorithms and methods I developed led to order-of-magnitude improvement in state-of-the-art results in important and long-standing theoretical physics problems (the Ising model). My publications can be found here. I am a very practiced speaker with extensive speaking experience. I have given over 50 academic seminars at universities all over the world including Harvard, Stanford, Berkeley, Princeton, and Cambridge, among others. I have also given several longer workshop style talks at international conferences as well as shorter invited talks at large prestigious conferences. I was an invited speaker at Strings 2015, the largest annual conference of the international String theory community (video is here).

Sylvain Zimmer: Developer-friendly task queues: what we learned building MRQ

Abstract:

MRQ is a gevent-based task queue, built with performance and developer productivity in mind. I will explain why we migrated away from Celery and then from RQ, explore some the unique features of MRQ and discuss the tradeoffs we made while designing it.

Speaker: Sylvain Zimmer (Pricing Assistant / dotConferences)

Speaker bio:

Sylvain Zimmer is a software developer and longtime free culture advocate. In 2004 he founded Jamendo, the largest Creative Commons music community online. Since 2012, he has been the CTO of Pricing Assistant, a startup specialized in large-scale crawling of E-commerce websites. He is also the founder and main curator of dotConferences, a series of TED-like developer events in Paris, as well as Paris.py, the Python meetup in Paris.

Thierry Parmentelat (INRIA), Olivier Ricou (EPITA) et Jill-Jênn Vie: Python MOOCs: Lessons Learned

Abstract:

Retour d’expérience sur les MOOC Python d’Inria et d’EPITA, deux approches complémentaires pour enseigner le langage à grande échelle. Table ronde animée par Jill-Jênn Vie.

Speaker: Thierry Parmentelat (INRIA), Olivier Ricou (EPITA) et Jill-Jênn Vie (RIKEN)

Speaker bio:

Thierry Parmentelat est ingénieur de recherche à Inria Sophia-Antipolis et utilise Python depuis plus de 10 ans pour ses travaux de recherche ainsi que pour le développement des plateformes expérimentales dont il a la charge. Il est l’auteur avec Arnaud Legout du MOOC Python : des fondamentaux jusqu’à l’utilisation du langage sur la plateforme edX France université numérique. Olivier Ricou est directeur de la recherche à EPITA et y enseigne l’analyse numérique, le calcul parallèle et la géopolitique de l’Internet. Il est auteur du MOOC Python pour les scientifiques sur la plateforme edX IonisX.

Thomas Moreau: Robustifying concurrent.futures

Abstract:

This talk presents in details concurrent.futures and its drawbacks. Then, it introduces loky, a library that robustifies concurrent.futures.ProcessPoolExecutor, and some of the major design choices.

Speaker: Thomas Moreau (CMLA - ENS Paris Saclay)

Speaker bio:

I am a PhD student at Centre de Mathématiques et de leurs applications (CMLA), ENS Paris-Saclay, since Fall 2014. My PhD subject is Automatic feature extraction for physiological time series. My research interests touch several areas of Machine Learning, Signal Processing and High-Dimensional Statistics. In particular, I am working on Convolutional Dictionary Learning, studying both their computational aspects and their possible application to pattern analysis. I am also studying on theoretical properties of learned optimization algorithms and their links to deep learning. I have been experimenting with different tools for parallel and distributed computation, from standard libraries like python multiprocessing to more advance tools such as CUDA or openMPI. I have been working on loky and the robustification of concurrent.futures for the past 2 years in collaboration with Olivier Grisel.

Vaibhav Singh: Machine Learning to moderate ads in real world classified’s business

Abstract:

In this talk we share our experiences on how we at OLX Berlin built machine learning models to moderate 100+ million classified ads every month. Audience will get a chance to experience a real world of content moderation and a race to beat online fraudsters and scammers.

Speaker: Vaibhav Singh (OLX Naspers Services GmbH)

Speaker bio:

Currently working as a Machine Learning Scientist, Content Moderation at OLX Berlin. Previously worked as Data Scientist in Credit Risk and Software Developer across Europe.

Venkat Raghav Rajagopalan: Introduction to pomegranate

Abstract:

pomegranate is a python module for probabilistic modelling focusing on both ease of use and speed, beating out competitors in benchmarks. In this talk I will describe how to use pomegranate to simply create sophisticated hidden Markov models, Bayesian Networks, General Mixture Models (and more!).

Speaker: Venkat Raghav Rajagopalan (Telecom Paristech)

Speaker bio:

I am a research engineer affiliated with Telecom Paristech, currently working from INRIA, Saclay under the Guidance of Dr. Alexandre Gramfort. I work on tree based methods of scikit-learn. Myself and Jacob collaborate a lot on the tree code base of scikit-learn. He introduced me to pomegranate. I’ve also closely followed the recent GMM overhaul of scikit-learn, which was done by our other scikit-learn developer Thierry. I would like to say people should listen to me because I have an h-index of over 200, can code in assembly, play in a rock band, know what “One Direction” is, have the best sense of humor but unfortunately none of these are true (evidently especially the last one ;( ) But I can promise to work on making the talk lively and interesting, with guidance from Jacob… :)

chticode@univ-lille1.fr: How to make teenage girls love coding using Python and the visual arts-oriented language Processing ?

Abstract:

Too few women work in IT. It’s a pity as it deprives women of exciting jobs and computing of women viewpoint. So, members of Lille University decided to fight prejudices by making female computer science students teach coding to teenage girls, using Python and the visual arts language Processing.

Speaker: chticode@univ-lille1.fr (Informatique au fénimin, Lille University)

Speaker bio:

Chti’code is a collective of 3 faculty members, Maude Pupin, Philippe Marquet and Yann Secq, teaching computer science at Lille University. They are all implied in popular science. For example, Yann Secq initiate teaching of coding for primary school pupils by computer science students ; Philippe Marquet argues for the teaching of coding from primary school in France and Maude Pupin promotes computer science for teenagers and, especially for girls with informatique au féminin collective. Together, they initiate L codent, L créent activity to : - introduce and demystify coding - prove that coding is accessible to any women - illustrate the creative power of coding in an artistic context - and, ideally, arouse vocations when orientation choices are made

harjinder.v2@gmail.com: DESIGNING AND CODING FOR CLOUD-NATIVE APPLICATIONS USING PYTHON

Abstract:

In this talk, the speakers share their experiences in designing and writing a cloud native machine learning application in Python. The attendees will learn to visualise different layers involved in cloud computing and will be able to write their cloud-native applications after attending this talk.

Speaker: harjinder.v2@gmail.com (RedHat)

Speaker bio:

Harjinder Mistry is currently a member of Developer-Tools team in RedHat, where he is incorporating data science into next-generation developer tools powered by Spark. Prior to RedHat, he was a member of IBM Analytics team and he developed Spark-ML pipeline components of IBM Analytics Platform. Earlier, he had spent several years in DB2 SQL Query Optimizer team building and fixing the mathematical model that decides the query execution plan. He holds M.Tech. degree from IIIT, Bangalore, India.

nicolas.chauvat@logilab.fr: Serve an Hypermedia API using CubicWeb-JSONSchema

Abstract:

Ever complained that you had to write client code specific to each RESTful APIs you use even though the services are similar ? Come learn about the use of HTTP, JSON Schema Hypermedia and JSON-LD to write more generic clients that do not depend on the specific URL scheme of an API.

Speaker: nicolas.chauvat@logilab.fr (Logilab)

Speaker bio:

Over 20 years of Python and still counting. Founder and CEO of Logilab. You do science or engineering ? Come learn Python with us !

Subscribe to Receive PyParis Updates

indicates required

Sponsors

Organiser

Silver