Nuno Castro

Data Science Leader


I'm a Data Science leader with 16 years experience in the field. I run the Machine Learning Solutions Lab UK team at Amazon Web Services. I previously headed the Personalization and Recommendations Data Science teams at Expedia Group, the Search and Sort data science teams at, and the global data science function for the Expedia Partner Solutions, through my 8 year tenure at Expedia Group. Before that, I was a Data Scientist at Feedzai, where I performed R&D in big data and machine learning, applied to large scale credit card fraud detection. I was also a Big Data Intern at Siemens Research in Princeton, NJ, where I implemented efficient search algorithms using Hadoop. I obtained a PhD in Machine Learning and Computer Science from the University of Minho, where I researched scalable machine learning algorithms for temporal data. Before, I was with Nokia Siemens as an R&D software engineer. I graduated in Computer Science and Systems Engineering in 2006.

  • Recent



    2022 - now
    Manager Applied Science @ Amazon Web Services

    Manager Applied Science

    Leading the UK ML Solutions Lab team, helping our customers optimize their business with ML

    View website

    2016 - 2022
    Director Data Science @ Expedia Group

    Director of Data Science

    Director ML Science for Personalization and Recommendations (4 teams, 22+ people). Director ML Science for Search and Sort, Head of Data Science for EPS, the B2B Expedia brand.

    View website

    2014 - 2015
    Senior Data Scientist @ Expedia Group

    Sr Data Scientist in the Analytics team

    Sr Data Scientist in the Analytics team of the world largest travel company.

    View website

    Data Scientist @ Feedzai, Inc.

    Data Scientist in a fraud prevention solution

    Research and development in a large scale credit card fraud prevention solution, which processes 2B credit card transactions a year. I have been working as a data scientist in the fraud detection classification tools. I have also led the development of a REST API web service for helping online merchants detect fraud in their payments.

    View website

    Summer 2012
    Big Data Intern @ Siemens Research

    Researcher in search and indexing techniques in big data

    Developed efficient search techniques in big data using Hadoop. The goal was to retrieve the Top-K nearest neighbors to the query sequence, for N queries at the same time. I also implemented a state of the art index for very fast approximate search.

    View website

    2007 - 2012
    Machine Learning PhD candidate @ University of Minho

    PhD in Machine Learning and Computer Science

    Research and Development of highly scalable pattern discovery algorithms for Terabyte sized disk-based or streaming data, and statistical evaluation measures for pattern discovery algorithms, published in top-tier conferences. One of the approaches won the Google-sponsored best student paper award and was also published in a journal.

    View website

    2006 - 2008
    R&D Software Engineer @ Siemens and Nokia Siemens

    Research and Development in a telecommunication networks analysis product

    Research and Development in a leading telecommunications network management and analysis product (SPOTS). The SPOTS is implemented in more than 90 countries at top telecommunication mobile operators (e.g. Vodafone,T-Mobile). After just 1 year, I was leading the online monitoring subsystem. This subsystem monitors thousands of network objects properties simultaneously, triggering alarm events in case anomalies are detected. I was also responsible for the product’s System Monitoring tool, and performed the research, analysis, and specification of Adaptive Thresholding features for the Real Time subsystem. Main technical skills covered: Java, C++, ClearCase, unix shell scripting, machine learning.

    2001 - 2006
    Computer Science degree @ University of Minho

    Computer Science and Systems Engineering degree

    During my 5 year undergraduate degree at the University of Minho I became well versed in programming in Java/C/C++/Perl/VB/SQL/PHP/HTML/Haskell/Prolog, artificial intelligence, machine learning, statistics, databases, object oriented programming, UML, software engineering, web programming, networks, data structures and algorithms, computation theory, cryptography, GUI design, XML/XPath/XSL, distributed programming, operating systems and computer architectures. My efforts were recognized by winning two university merit awards in 2003 and 2004. I spent a semester abroad at the Utrecht University (The Netherlands), where I attended courses and performed projects supervised by Doaitse Swierstra. I spent my last semester performing research in event forecasting at Siemens as an intern, where my project achieved a grade of 19/20.


    “How Expedia is using deep learning to improve the hotel booking process”, in Phocuswire

    Nuno Castro

    Expedia Partner Solutions is using deep learning models to improve the hotel booking process for partners and - ultimately - the end consumer.

    Full article.

    “How Machine Learning Can Drive Your Business’ Success Today ”, in ITB Berlin 2019

    Nuno Castro

    Nuno Castro speaking at #ITBBerlin “Through EPS’s API, partners can unlock the power of Expedia Group and get access to our machine learning” #travel #epsrapid #api

    Machine learning is something we hear a lot about – we know it’s going to make our lives easier and our businesses more successful but how can you implement it in your businesses today? Learn how machine learning can easily be part of your offering – no data science skills needed!

    Linkedin Link.

    “How to get a job as a data scientist: Qualifications & skills”, in Computerworld UK

    Nuno Castro

    With the rise of big data comes the need for more highly skilled people to mine and interpret that data for businesses. This is the role of a data scientist, the job that Harvard Business Review called "the sexiest job of the 21st century" back in 2012.

    View article

    “Ranking Hotel Images using Deep Learning”, in PyData 2017

    Nuno Castro

    PyData London 2017 talk where I cover how Expedia has been ranking hotel images using deep learning.

    Presentation video.

    “A deep learning chatbot for hotel search”, in Big Data & Analytics Innovation Summit

    Nuno Castro

    "Chatbots have recently received a great deal of attention from both industry and academia. Cutting edge advances in deep learning (namely LSTM) have enabled these developments. In this talk, we will share how Expedia Affiliate Network is using a chatbot built using deep learning to manage a hotel search conversation with consumers."

    Presentation video.

    “Room with a View: How Expedia Will Pick the Right Hotel Photos for You”, in NVIDIA Blog

    Nuno Castro

    "We all need to make a good first impression. Even hotels. That’s why travel giant Expedia is using AI to help hotels put the right photos in front of the right people. Travelers spend less than a second deciding what they think of a place. So those pictures better be Instagram-worthy..."

    Read more.

    “Significant Motifs in Time Series”, in Statistical Analysis and Data Mining

    Nuno Constantino Castro and Paulo J. Azevedo

    Extends the co-winner of the Google sponsored best student paper award from SDM’11. This paper gives a method to evaluate statistical significance of discovered patterns in time series, enabling ranking and filtering of the often large number of patterns discovered by time series data mining techniques. In addition to additional details on the algorithms, this paper includes additional results.

    View website

    “Time Series Motifs Statistical Significance”, in SDM'11, Phoenix, AZ

    Nuno Constantino Castro and Paulo J. Azevedo

    An approach for assessing (for the first time in the literature) the statistical significance of time series patterns. Statistical significance tests are used to assess each pattern’s p-value. [Best Student Paper].

    View Website

    “Multiresolution Motif Discovery in Time Series, in SDM'10, Columbus, OH

    Multiresolution Motif Discovery in Time Series

    A highly efficient algorithm for pattern discovery in time series data. The algorithm finds all patterns in the database in linear time: uses one single sequential scan over the database; and allows adjusting the amount of memory to use using a clever space saving approach.

    View website


    Feel free to drop me a line.

    I'm based in London.