Register now

29 September 2020 14:00-18:00

We are back!

And we certainly hope our timing is better this time. After having to cancel our large in-person conference with 300 participants in March due to a certain virus, we have now made the decision to run the conference as an online event instead.

The agenda and content are mostly the same as what was planned for March, meaning we have a line-up of exciting speakers from industry and academia who will share their experience and knowledge of everything from attribution modeling and machine learning pipelines to social robots and data ops.

As usual, our events are organized by female data scientists and feature only women on stage. However, participants of all genders and backgrounds are more than welcome to attend! Presentations are intentionally fairly technical and aimed towards current or aspiring data scientists, data engineers, machine learning engineers, as well as other AI experts.

Invited Speakers

Farnaz Motamediyan
Co-Founder and CEO of queensai.com

Applying Advanced Statistics for Attribution Models: When input data is the biggest challenge

This talk discusses how to apply advanced statistics for complex problems, in a simple way. An example complex problem is, attributing credit to different channels based on their continuation to user acquisition. This talk will show how game theory and probability theory can be applied to make a model that could fairly represent the user journey prior to their registration.

Farnaz Motamediyan

Farnaz Motamediyan is the co-founder and CEO of queensai, an education technology platform for peer-to-peer learning and mentorship in tech. Farnaz is a Data-Scientist by profession and a proud ex.iZettler. She is also a board member of the PinkProgramming organization that aims to inspire more women to enter the IT industry. Her background in both Computer Science and Innovation Technology has brought her to the current journey of leveraging technology and AI for education and learning.

Read more

Josefin Scott
Lead Software Architect at Recorded Future

Deduction from Production

When developing models we spend a significant amount of time evaluating performance, simulating real distributions, and trying to avoid biases. But how do we ensure that we are getting the expected results after release? While looking at concrete examples, we will explore the evaluation and monitoring of deployed models, what to consider before deployment, and what to do when things don’t go as planned.

Josefin Scott

Josefin Scott has a Master’s degree in Computer Science (algorithms, languages, and logic) from the Chalmers University of Technology in Gothenburg. At Recorded Future, she focuses on the design and implementation of new machine learning strategies for natural language processing (NLP). She is also a co-founder and board member of the non-profit organization, Gothenburg Artificial Intelligence Alliance.

Read more

Iolanda Leite
Assistant Professor at KTH Royal Institute of Technology

Human-centered AI: challenges and opportunities

As AI systems become ubiquitous in people’s lives, a long-standing barrier is a need for these systems to properly respond and adapt to people and to the complex social dynamics of different environments. In this talk, I will present current research on using AI to enable robots with the social capabilities that enable them to engage with people in real-world environments. I will also discuss limitations and opportunities of the state of the art of AI, arguing that an improved understanding of how these systems perceive, reason and act depending on their surrounding social context can lead to more natural and efficient interactions with people.

Iolanda Leite

Iolanda Leite is an Assistant Professor at the School of Electrical Engineering and Computer Science at KTH Royal Institute of Technology. She holds a Ph.D. in Information Systems and Computer Engineering from IST, University of Lisbon. Prior to joining KTH, she had postdoctoral appointments at Yale University and at Disney Research. Iolanda's research interests are in the areas of artificial intelligence and human-robot interaction. She is the co-editor in chief of AI Matters, the newsletter of the ACM Special Interest Group in Artificial Intelligence (SIGAI).

Read more

How to join

Once you have registered for this event, you will receive a confirmation email containing a link to a Zoom webinar. This link will become active on the day of the event, and you will also need to enter the password shared with you over email to access the live stream.

In addition to this, you will also be able to see the joining information (i.e. Zoom link and password) by returning to this page after registering and clicking the "joining details" section at the top of this page.

Contact us at women@wids.se if you have trouble accessing the live stream!

Register now

Schedule

14:00

Introduction

14:20

Human-centered AI: challenges and opportunities

As AI systems become ubiquitous in people’s lives, a long-standing barrier is a need for these systems to properly respond and adapt to people and to the complex social dynamics of different environments. In this talk, I will present current research on using AI to enable robots with the social capabilities that enable them to engage with people in real-world environments. I will also discuss limitations and opportunities of the state of the art of AI, arguing that an improved understanding of how these systems perceive, reason and act depending on their surrounding social context can lead to more natural and efficient interactions with people.


Iolanda Leite, Assistant Professor at KTH Royal Institute of Technology

Read more
Iolanda Leite
Assistant Professor at KTH Royal Institute of Technology
14:50

Deduction from Production

When developing models we spend a significant amount of time evaluating performance, simulating real distributions, and trying to avoid biases. But how do we ensure that we are getting the expected results after release? While looking at concrete examples, we will explore the evaluation and monitoring of deployed models, what to consider before deployment, and what to do when things don’t go as planned.


Josefin Scott, Lead Software Architect at Recorded Future

Read more
Josefin Scott
Lead Software Architect at Recorded Future
15:20

Break

15:40

Applying Advanced Statistics for Attribution Models: When input data is the biggest challenge

This talk discusses how to apply advanced statistics for complex problems, in a simple way. An example complex problem is, attributing credit to different channels based on their continuation to user acquisition. This talk will show how game theory and probability theory can be applied to make a model that could fairly represent the user journey prior to their registration.

Farnaz Motamediyan, Co-Founder and CEO of queensai.com

Read more
Farnaz Motamediyan
Co-Founder and CEO of queensai.com
16:10

Lightning Talks: Machine Learning Tools and Data Infrastructure in everyday work

  • Maria Henningsson, Data Scientist at IKEA: A data science tech stack for empowerment – spreading insights to the many
  • Kristina Kosenko, Founder Data Scientist at Daxoni: ML Infrastructure in Health Industry
  • Alva Liu, Machine Learning Engineer at Spotify: ML infrastructure @ Spotify Search
  • Alice Karnsund, AI Engineer at King: AI Powered Game Content Production
  • Satya Vardhini, Data Engineer at Tele2: Operationalising Machine Learning @Tele2
  • Sarah Hantosi Albertsson, Data Scientist at Scania: DataOps at Scania
Read more
17:10

Closing Remarks

17:20

Virtual Mingle

Lightning Talk Speakers

Kristina Kosenko
Founder Data Scientist at Daxoni

ML Infrastructure in health-tech

Lots of healthcare companies are dealing with data ingestion from multiple external data sources to feed their models. Those sources could be clinics, insurance companies, labs and different medical devices (heart monitors, glucose meters etc.)

This comes with a great deal of effort for defining general schema, data processing and mapping. When we are dealing with patients personal and medical information, not only pipeline effectiveness, but data quality and data privacy needs to be considered.

In a lot of of cases, pipelines that include multiple lamba functions, redshift/presto/hive together with S3 and EMR and Sagemaker are used.

Kristina Kosenko

I am a data scientist and founder at Daxoni. I come from a world of computer science and applied linguistics. Right after university, I joined the world on data science and now it been almost 7 years. Throughout these years I have been on hands data scientist, was leading teams, been a member of the steering committee and tech squad teams. Also I was a founding partner of a collaboration platform for data scientists, that allowed people to learn and share knowledge. Majority of my experience is in health-tech, but i have been working with fintech, media and IP. At everyday work day I do all kind of stuff, from data warehousing with Redshift, model building on pagemaker to visualization in Looker, but I would say my favorite would be anything connected with messy unstructured text.

Read more

Sarah Hantosi Albertsson
Data Scientist at Scania

DataOps at Scania

At the geospatial intelligence team at Scania we build continuous delivery pipelines that enable us to iteratively improve our code, our algorithms and data products. Pipelines that transform the data of our global connected fleet of vehicles into data products we can use to serve a range of end goals. Today our pipes run on a Hortonworks platform using Apache Spark Streaming and the data products are pushed to Apache Kafka. In the build pipeline we use Jenkins, Artifactory and Ansible to test, deploy and run the jobs. I will show you how we use these technologies to ensure quality, repeatability and scalability for our pipelines and data products.

Sarah Hantosi Albertsson

Sarah works as a data scientist in the Connected Intelligence team at Scania's R&D section. A team who create innovative and scalable data products to ensure actionable insights. Together with her team she finds the technical architecture and implementation to realise business opportunities. Whether that's implementing big data pipelines, tweaking Kafka throughput, optimising algorithms utilising Spark or building data visualisations for logistics optimisation. With a degree in Cognitive Science she is at her best where she can combine her interest in people, service design and technology. Where cross-functional teams and working end-to-end is part of her daily routine.

Read more

Satya Vardhini
Data Engineer at Tele2

Operationalising Machine Learning @Tele2

The big picture of how Tele2, as a telecom company, leverages Machine Learning. We will  walk through our ML life cycle using one of our use cases as an example. Technical details of how we use Snowflake, Gitlab CI, Airflow, AWS S3 and AWS SageMaker to build and productionalize ML models.

Satya Vardhini

Satya is a Data Engineer at the BI and Analytics department at Tele2. She works in a team that builds and delivers production scale machine learning systems that impact Tele2's business functions. Prior to that she worked as a Software Engineer at Ericsson enabling CI/CD for Radio Products. She holds a master's degree in computer science and has done her thesis working with convolutional neural networks.

Read more

Alice Karnsund
AI Engineer at King

AI Powered Game Content Production

Talk about research and development platform for AI powered game-content production. It is a good example of how you can take AI research to an ML pipeline. The platform serves as a tool for centralised analysis of models and inference, and has a standardised pipeline for training and inference. It can easily be extended to new games and different types of models, and provides game designers and data scientists with valuable insights of how to tweak the content in order to improve the games.

Alice Karnsund

Alice Karnsund has a background in Engineering Physics with a Master’s degree in Machine Learning from KTH Royal Institute of Technology. As part of the AI R&D team at King, her main focus is within ML-driven content production and applying Reinforcement Learning for content testing.

Read more

Alva Liu
Machine Learning Engineer at Spotify

ML infrastructure @ Spotify Search

Search is crucial to the Spotify experience. It acts as the entry point to Spotify's expansive audio catalog, allowing users to explore music and podcasts and find exactly what they're looking for. Spotify search wouldn't be what it is without Searchrank, the machine learning-powered result re-ranker. Searchrank provides personalized search results to all Spotify users, serving tens of thousands of requests per second at very low (<20ms) latency. In this talk, we will go into the infrastructure around the search ranking service, the requirements that lead us to design the service the way we did, especially focusing on challenges due to the low latency requirements.

Alva Liu

Alva is a ML Engineer in the Search product area at Spotify. Her team in Search mainly focuses on result quality and user intent understanding, and owns multiple large-scale ML systems for Spotify Search. She has a master’s degree in Machine Learning from KTH and has been working at Spotify since 2019.

Read more

Maria Henningsson
Data Scientist at IKEA

A data science tech stack for empowerment – spreading insights to the many

At IKEA, we believe that data is key to realising our vision, to create a better everyday life for the many people. To make the greatest impact as data scientists, we need streamlined tools to put the right data products and insights directly into the hands of our 166,000+ coworkers.We use Google Cloud Platform’s suite of tools for data storage, compute, and machine learning, but we also build our own custom solutions on top of that – to achieve scale, flexibility, and efficient use of resources. Fundamentally, it’s all about making sure the data science tech stack enables us data scientists to support the business delivering value to the customers. State-of-the-art technology - in the service of people.

Maria Henningsson

Maria Henningsson has an MSc in engineering mathematics and a PhD in Automatic Control, both from Lund University. She has 10+ years of experience working with mathematical modeling in various different industries, from automotive and aerospace to telecom and retail. When she’s not working, she is in avid reader and passionate about learning more on technology, people, and the world in general.

Read more

Organized by