MACHINE LEARNING
IN PRODUCTION

Technical Events For Technical People

TECH SESSIONS
Thursday, February 8th
18:00 – 21:30

 

In Partnership With

Stay Updated

MACHINE LEARNING
IN PRODUCTION

Technical Events For Technical People

TECH SESSIONS
Thursday, February 8th
18:00 – 21:30

 

In Partnership With

Stay Updated

A technical event for technical people.

FOUR INDUSTRY CASE STUDIES
ONE ML START-UP PITCH
KEY NETWORKING
LOTS OF CONTENT

This is a learning event. No high-level vagueness, just powerful insights for you.

Themes:
{Machine Learning} {Automation} {Articial Intelligence} {Predictive Analytics}

A technical event for technical people.

FOUR INDUSTRY CASE STUDIES
ONE ML START-UP PITCH
KEY NETWORKING
LOTS OF CONTENT

This is a learning event. No high-level vagueness, just powerful insights for you.

Themes:
{Machine Learning} {Automation} {Articial Intelligence} {Predictive Analytics}

Event Schedule

18:00
Event Check-In & Networking


———

18:30 – 18:50
Building Multilingual Recommendations Systems for BBC News

Magda Piatkowska, Lead Data Scientist, BBC
Magda Piatkowska, Lead Data Scientist, BBC

Key Takeaways

Talk:
Introduction to a high level architecture of recommendation system based on multilayer, component based structure.
To build a production system we are starting simple. But it is not only about the technical challenge related to scale, but it is also about the journalist behind it, the audience in front of it and the data scientist in the middle.
But to design that system we had to start blind. Therefore complex, unsupervised methods to deal with the scale were necessary.


———

18:50 – 19:10
Building Robust Machine Learning Systems

Stephen Whitworth, Ravelin
Stephen Whitworth, Co-Founder & Machine Learning Engineer, Ravelin


———

19:20 – 19:35
Networking Break


———

19:35 – 19:55
Bayesian Online Change-Point Detection at Scale

Paolo Puggioni, Schroders
Paolo Puggioni, Machine Learning Data Scientist, Schroders

Key Takeaways

Talk:
– Online change-point detection in time series.
– Handling different types of distributions.
– Running it at scale.
– Automated generation of alerts.


———

19:55 – 20:15
How We Enabled Faster Iteration on Recommendation Algorithms

Saul Vargas, ASOS
Saúl Vargas, Senior Data Scientist, ASOS

Key Takeaways

Talk:
– ASOS Data Science team journey from a rigid Spark-based solution to a flexible Keras-powered recommendation algorithm.
– We outline the limitations of the original approach in the context of large-scale, fashion recommendations.
– Keras as a great tool to build not-so-deep recommendation algorithms based on factorisation models.
– How we are quickly adding new features to our model: metadata and image features, time and location awareness, etc.


———

20:15 – 20:25
Modeling the Importance of Flight Partners at Skyscanner

Tatia Engelmore, Skyscanner
Tatia Engelmore, Data Science Manager, Skyscanner

Key Takeaways

Talk:
Skyscanner works with a wide variety of flights partners (airlines and online travel agents). Currently all partners who provide the selected flight are shown when a traveler goes to book. We would like to better understand the implications of losing coverage for a partner: will travelers still find the flights they’re looking for? A strategy for modeling the impact of changing the partners offered will be presented.


———

20:25 – 20:35
Constructing Flight Itineraries with Machine Learning

Dima Karamshuk, Skyscanner
Dima Karamshuk, Senior Data Scientist, Skyscanner

Key Takeaways

Talk:
As a leading travel marketplace, Skyscanner is serving a daily load of up to a dozen billion flight itineraries to its users across the globe. The distribution of travel quotes at such a scale requires efficient search algorithms for maximizing the relevance and comprehensiveness of the itineraries to the travelers while minimizing the load on the partners (airlines and travel agencies). This talk will be focused on using machine learning for optimizing dynamic content distribution at scale and will shed the light on Skyscanner’s efforts in this direction.


———

20:35 – 21:30
Networking & Drinks

This is your chance to meet the speakers and your fellow machine learning enthusiasts, making connections that go well beyond this evening!

Event Schedule

Check-In & Networking
18:00

Magda Piatkowska, Lead Data Scientist, BBC

Building Multilingual Recommendations Systems for BBC News
Magda Piatkowska, Lead Data Scientist, BBC
18:30 – 18:50
LinkedIn

Key Takeaways

Talk: Introduction to a high level architecture of recommendation system based on multilayer, component based structure.
To build a production system we are starting simple. But it is not only about the technical challenge related to scale, but it is also about the journalist behind it, the audience in front of it and the data scientist in the middle.
But to design that system we had to start blind. Therefore complex, unsupervised methods to deal with the scale were necessary.

Stephen Whitworth, Ravelin

Building Robust Machine Learning Systems
Stephen Whitworth, Co-Founder & Machine Learning Engineer, Ravelin
18:50 – 19:10
LinkedIn

Networking Break
19:20 – 19:35

Paolo Puggioni, Schroders

Bayesian Online Change-Point Detection at Scale
Paolo Puggioni, Machine Learning Data Scientist, Schroders
19:35 – 19:55
LinkedIn

Key Takeaways

Talk:
– Online change-point detection in time series.
– Handling different types of distributions.
– Running it at scale.
– Automated generation of alerts.

Saul Vargas, ASOS

How We Enabled Faster Iteration on Recommendation Algorithms
Saúl Vargas, Senior Data Scientist, ASOS
19:55 – 20:15
LinkedIn

Key Takeaways

Talk:
– ASOS Data Science team journey from a rigid Spark-based solution to a flexible Keras-powered recommendation algorithm.
– We outline the limitations of the original approach in the context of large-scale, fashion recommendations.
– Keras as a great tool to build not-so-deep recommendation algorithms based on factorisation models.
– How we are quickly adding new features to our model: metadata and image features, time and location awareness, etc.

Tatia Engelmore, Skyscanner

Modeling the Importance of Flight Partners at Skyscanner
Tatia Engelmore, Data Science Manager, Skyscanner
20:15 – 20:25

Key Takeaways

Talk: Skyscanner works with a wide variety of flights partners (airlines and online travel agents). Currently all partners who provide the selected flight are shown when a traveler goes to book. We would like to better understand the implications of losing coverage for a partner: will travelers still find the flights they’re looking for? A strategy for modeling the impact of changing the partners offered will be presented.

Dima Karamshuk, Skyscanner

Constructing Flight Itineraries with Machine Learning
Dima Karamshu, Senior Data Scientist, Skyscanner
20:25 – 20:35

Key Takeaways

Talk: As a leading travel marketplace, Skyscanner is serving a daily load of up to a dozen billion flight itineraries to its users across the globe. The distribution of travel quotes at such a scale requires efficient search algorithms for maximizing the relevance and comprehensiveness of the itineraries to the travelers while minimizing the load on the partners (airlines and travel agencies). This talk will be focused on using machine learning for optimizing dynamic content distribution at scale and will shed the light on Skyscanner’s efforts in this direction.

Networking & Drinks
20:35 – 21:30

This is your chance to meet the speakers and your fellow machine learning enthusiasts, making connections that go well beyond this evening!

Ticket Registration

Previous Content

Content: Machine Learning in Production

Continuous Machine Learning Model Deployment
Stephen Whitworth, Co-Founder
Ravelin

Bayesian Change-Point Detection
Paolo Puggioni, ML Data Scientist
Schroders

Next up:
MARKETING DATA SCIENCE

A Technical Event For Technical People

A Technical Event
For Technical People

Thursday, April 12th
18:00 – 21:30

 

In Partnership With
Facebook Logo

Join Our Next Event

Next up:
MARKETING DATA SCIENCE

Technical Events For Technical People

A Technical Event For Technical People
Thursday, April 12th
18:00 – 21:30

 

In Partnership With

Join Our Next Event