• 08:00

    REGISTRATION & LIGHT BREAKFAST

  • 09:00
    Dino Bernicchi

    WELCOME NOTE

    Dino Bernicchi - Data Science Strategy Consultant -

    Down arrow blue

    Using Computer Vision with Recommender Systems to Transform Retail

    There has been an explosion in the number of e-commerce stores, available products, and well-crafted content. Recommender Systems have therefore come to the forefront in attracting customers, matching them to products they will love, and to content they will find engaging. But what happens when you don’t have many data points on your customers or products – or worse – customers are browsing your online store anonymously. How do you still deliver an engaging experience that drives your bottom line? Join this talk to discover how Computer Vision, Natural Language Processing, and other AI technology is being used with Recommender Systems to transform retail.

    Dino Bernicchi has 12 years’ experience driving AI strategy and developing productionized data solutions for large South African corporates such as TFG, Woolworths, Standard Bank, and Chevron SA. Most recently, as Head of Machine Learning at Homechoice - a home-shopping and financial services group - a successful AI strategy has driven R140m in additional operating profit at 15x return on investment. Dino is a regular speaker at AI, and data conferences, and lectures the Data Science Leadership for Executives course at University of Cape Town's Graduate School of Business. He also consulted on the development of course content for Andrew Ng's DeepLearning.AI.

    Linkedin
  • INDUSTRY INSIGHTS

  • 09:15
    Stephen Jordan

    AI in Payments Risk: Insights From a 0-to-1 Machine Learning initiative

    Stephen Jordan - Senior Data Scientist - Shopify

    Down arrow blue

    AI in Payments Risk: Insights from a 0-to-1 Machine Learning initiative

    Every month, Shopify Payments processes billions of dollars of sales for our merchants worldwide. At this scale, Shopify takes on sizable financial risk when shops face a surge in refunds and chargebacks but lack the funds to cover these charges. To address this challenge, we have started using machine learning to safeguard against losses on our Payments services. In this talk, I’ll share how we took our model from ideation to deployment, focusing on the technical challenges encountered and how we iterated quickly in a complex problem space.

    Stephen is a Senior Data Scientist at Shopify, working on risk management for the company’s financial products. He is interested in solving complex problems with data, aggressive automation, and mentoring others. Past work experience covers data science and engineering in the domains of fraud, credit risk, NLP for customer understanding, product growth and more. Stephen also has several patent applications pending for novel machine learning applications in user experience workflows.

    Linkedin
  • 09:40
    Sunando Das

    Machine Learning Enabling the Anatomy of Retail Media Measurement and Optimisation

    Sunando Das - Global Director - Predictive Analytics - Unilever

    Down arrow blue

    Machine Learning Enabling the Anatomy of Retail Media Measurement and Optimisation

    Forrester has predicted that the retail media market will more than double in the next four years to a $100bn+ market. However, the world federation of advertisers found (2021 report) that measurement in this space was one of the key capability gaps in marketing. Machine Learning (ML) has a significant role to play to close this capability gap. This presentation will outline how different ML techniques (such as ensemble models, and deep learning neural networks) can help solve the measurement and optimisation challenges of driving short-term and long-term returns on investments in retail media spend as well as improving the basis of activation.

    Sunando Das leads the Predictive Marketing and Shopping Analytics global centre of excellence in the Consumer and Market Insight team at Unilever. As part of his role, Sunando works with a team of data scientists, data engineers, econometricians, insights specialists and a network of specialist agency partners to create data science/ advanced analytics solutions, deploy scalable global programs across markets and drive business outcomes with stakeholders across functions. An accomplished speaker and thought leader, Sunando regularly publishes/ presents in platforms such as AI Summit, CIOapplications, Ad:Tech, AdMap, ARF Re:Think, Campaign, ESOMAR, Global Service Quality, Journal of Brand Management, i-Media. His work on AI algorithms for personalization strategies won the Research-Live Innovation of the year 2015 award, use of machine learning to uncover innovation platforms won MRS 2016 award for impactful insights. Sunando is a veteran data science practitioner with two decades of experience across strategic insights, adtech/ martech and consulting. Sunando is an MBA graduate with B.S. in Statistics. He has Data Science IoT certificate from the University of Oxford.

    Linkedin
  • INFRASTRUCTURES & FRAMEWORKS

  • 10:05
    Florentin Kristen

    An NLP Inspired Deep Recommender Engine For Out-of-Catalog Items

    Florentin Kristen - Machine Learning Engineer - Chrono24

    Down arrow blue

    An NLP inspired deep recommender engine for out-of-catalog items

    At Chrono24 we are dealing with the largest inventory of luxury watches worldwide, covering >10 million of watches. Many of these items are limited editions or in other ways unique. In this talk we want to recap the journey from using a collaborative filtering approach for our watch recommender system to a modern, custom deep learning approach. We show how we solved the problem to recommend unseen not-yet-in-catalog watches, by building an embedding model based on their attributes, using techniques from natural language processing.

    Florentin Kristen is working as a Machine Learning Engineer at Chono24 GmbH since 2020. He is focused on using machine learning to build an exclusive relationships with the customers, provide a safe shopping experience and increase the marketplace quality overall. Florentin implements and deploys computer vision and NLP models, including image quality assessment, image classification and recommender systems. Before joining Chrono24 GmbH he finished his degree in Computer Science at KIT(Karlsruhe Institute of Technology) with a specialization on deep learning and computer vision.

    Linkedin
  • Christian Freischlag

    An NLP Inspired Deep Recommender Engine For Out-of-Catalog Items

    Christian Freischlag - Lead Machine Learning Engineer - Chrono24

    Down arrow blue

    An NLP inspired deep recommender engine for out-of-catalog items

    At Chrono24 we are dealing with the largest inventory of luxury watches worldwide, covering >10 million of watches. Many of these items are limited editions or in other ways unique. In this talk we want to recap the journey from using a collaborative filtering approach for our watch recommender system to a modern, custom deep learning approach. We show how we solved the problem to recommend unseen not-yet-in-catalog watches, by building an embedding model based on their attributes, using techniques from natural language processing.

    Christian Freischlag is Lead Machine Learning Engineer at Chrono24 since 2016. Christian's focus is developing systems based on Deep Learning and operating them according to ML-Ops principles for use cases like NLP-based fraud-detection, watch (image) classification and recommender systems. Before joining Chrono24, he was a Big Data Engineer at Connexity (now Taboola) and worked in data engineering at CERN while finishing his Master's in Computer Science.

    Linkedin
  • 10:30

    COFFEE & NETWORKING BREAK

  • 11:05
    Clara Castellanos Lopez

    Improving Property Content and Traveler Experience with ML

    Clara Castellanos Lopez - Senior Machine Learning Scientist - Expedia Group

    Down arrow blue

    Improving Property Content and Traveler Experience with ML

    At Expedia, our mission is to power global travel for everyone, everywhere. Presenting high quality content about properties all over the world, whether it is a hotel or vacation rental, in over 40 languages, requires collaboration between humans and algorithms. For example, displaying suitable images or relevant amenity information allows travelers to understand their booking, assess whether it meets the expectations and ultimately have the best travel experience. In this talk I will share some ways in which Expedia is leveraging machine learning to enrich and improve property content.

    Clara Castellanos Lopez is a Senior Machine Learning Scientist at Expedia, previously, she was Senior Data Scientist at QBE Insurance Group where she worked with the claims teams building machine learning algorithms to provide data driven insights and automated solutions. Clara has been working in the industry since 2014 with experience in oil and gas, retail and insurance. She has a master degree in Applied Mathematics and a Ph.D in Geophysics from Universite de Nice Cote d’Azur.

    Twitter Linkedin
  • 11:30
    Anirudh Sanga

    Applied Intelligence: A Scalable Path to Production for Your AI Initiatives

    Anirudh Sanga - Enterprise Lead - Palantir

    Down arrow blue

    Applied Intelligence: A Scalable Path to Production for Your AI Initiatives

    Integrating AI seamlessly into operations requires an architecture that allows Tech, Analytics, and Business teams to be both producers and consumers of data. In this talk, Anirudh will talk about Palantir’s approach to operationalising AI through a common data platform where every part of the business can work together to build operational AI. This will be brought to life with a Dynamic Pricing application demo used by a major retailer to execute their pricing strategies on 100k+ SKUs in real-time.

    Anirudh is a technical lead at Palantir with experience in Software Engineering, Product Management, and Operations Optimization. He leads cross-functional teams at client engagements across across Europe, North America and APAC, generating value with Technology and Data. He is driven by work at the intersection of Engineering, Leadership and Social Impact.

    Linkedin
  • 11:55

    Panel: The Rise of Automation: How Will Retail be Impacted?

  • Merve Alanyali

    MODERATOR

    Merve Alanyali - Head of Data Science Academic Partnerships and Research - LV= General Insurance

    Down arrow blue

    Input data quality control with LV=’s Input-Checker

    During the production of machine learning models, the quality of the data being received is crucial to return precise and trustworthy predictions. Without thorough checks in place, there is no end to the number of erroneous data points that could be passed by the system that the models are deployed to. But how can our model sniff out these issues? LV=’s input-checker offers an answer to these questions. In my talk, I will be introducing our python package input-checker; a light weight open source tool providing data checks at the point of inference.

    Dr Merve Alanyali is Head of Data Science Academic Partnerships and Research at LV= General Insurance drawing on an interdisciplinary background in computer science, complex systems and behavioural science. Merve has a PhD in Data Science from University of Warwick, fully-funded by the Chancellor’s International Scholarship. Her research at The Alan Turing Institute and University of Warwick focuses on analysing large open data sources with the concepts from image analysis to machine learning to understand and predict human behaviour at a global scale. The examples include identifying protest outbreaks using Flickr pictures, estimating household income with Instagram pictures and predicting non-emergency incidents in New York City. Her work has received more than 100 citations and featured on television and press worldwide including coverage in the Financial Times and Bloomberg Business. Prior to her PhD, she was awarded a double Masters degree in Complex Systems Science by the University of Warwick, UK and Chalmers University of Technology, Sweden.

    Twitter Linkedin
  • Graeme Blyth

    PANELIST

    Graeme Blyth - Machine Learning Software Engineer - Popsa

    Down arrow blue

    Graeme is a Machine Learning Software Engineer at Popsa, where he has been at the forefront of utilizing cutting-edge ML techniques and products. At Popsa, Graeme has written the algorithms behind the Suggested Albums feature, collaborated on a novel face clustering algorithm, and led the development of a customer revenue prediction model using minimal in-app behaviour data.

  • Arslan Awan

    PANELIST

    Arslan Awan - Senior Data Science Engineer - Tesco Bank

    Down arrow blue

    An AI Scientist with an interest in computer vision and NLP, Arslan recently completed an MPhil in Artificial Intellingence (for electric vehicles) at University of Warwick while working part time as a Teaching Assistant. 3+ years experience in ML with a track record in building successful algorithms and predictive models for different use cases. Highly adept in predictive modelling, computer vision, web scraping, data analysis and visualisation with a specialisation in deep learning libraries such as Tensorflow. Passionate candidate with experience in applying CNNs to real world data.

    Linkedin
  • 12:40

    LUNCH

  • 13:50
    Marion Riffaud

    Would’ve Gone to Specsavers… Catching Up On Missed Appointments Using ML

    Marion Riffaud - Data Scientist - Specsavers

    Down arrow blue

    Would’ve gone to Specsavers… Catching up on missed appointments using machine learning

    Routine sight tests are essential for optometrists to identify and prevent health issues that can lead to sight loss, such as glaucoma. The first COVID-19 lockdown did not only prevent patients from having their routine checks but also created a backlog of overdue appointments. Join Marion Riffaud, Data Scientist at Specsavers, to learn how they implemented a machine learning solution to support CRM campaign managers in detecting customers that were expected to have booked an appointment since the start of the pandemic.

    Marion is a data scientist that has been working at Specsavers since 2020. During these challenging times, she helped the business on its data transformation journey and supported the growth of the data science team. She has a background in marketing and has been working in the retail industry for the last 6 years.

  • TREND PREDICTION

  • 14:15
    Ioannis Kogias

    Predicting Churn Before It Happens

    Ioannis Kogias - Lead Data Scientist - Dunelm

    Down arrow blue

    Predicting Churn Before It Happens

    Keeping customers engaged for longer helps build purchasing habits that can significantly increase their lifetime value. But how can we identify disengaged customers before it’s too late? At Dunelm, a cross-functional team came together to build & deploy a machine learning model to identify customers at risk of churning, and to successfully integrate it into existing CRM processes to optimise & personalise customer targeting. In this talk, I will walk you through the journey of what it took to make this project a success; from a technical standpoint the considerations we faced when building such a model, to the level of close collaboration that was required between business, data engineering, and data science teams.

    Ioannis is a Lead Data Scientist at Dunelm, with a background and PhD in Quantum Information. His advanced analytics & consultancy experience spans 5 years across various industries including Retail, FMCG and Oil & Gas, with experience on propensity modelling, timeseries forecasting, and predictive maintenance, among others. Ioannis led the development of an ML project that won the DataIQ Awards 2020 - Most innovative use of AI. He is currently developing Dunelm’s first data science products.

    Linkedin
  • 14:40
    Nicolas Peltre

    Implementing a Markov Chains-based Multi-Touch Attribution Model at BlaBlaCar

    Nicolas Peltre - Data Scientist - BlaBlaCar

    Down arrow blue

    Implementing a Markov Chains-based Multi-Touch attribution model at BlaBlaCar

    Marketing attribution models are used to attribute conversions to marketing channels and hence monitor the performance of online marketing investments. Rules-based attribution models are frequently used in the industry with models such as ‘last interaction’, ‘first interaction’ or ‘position based’. They all follow a simple set of rules to perform the attribution, however the rules are set arbitrarily and may therefore be biased. At BlaBlaCar, we developed a data-driven Multi-Touch attribution model based on Markov chains. We will present the challenges that we have overcome from the idea to the roll-out and the impact on performance marketing of this project.

    Nicolas is a Data Scientist at BlaBlaCar specialized in marketing topics: he is currently developing a Multi-Touch attribution model based on Markov Chains and participates in carrying out an incrementality framework to evaluate the performance of marketing channels. He previously worked for an insurance company developing end-to-end NLP models to automate and ease customer services tasks. He holds a MSc from CentraleSupélec.

    Linkedin
  • Ken Morange

    Implementing a Markov Chains-based Multi-Touch Attribution Model at BlaBlaCar

    Ken Morange - Data Analyst - BlaBlaCar

    Down arrow blue

    Implementing a Markov Chains-based Multi-Touch attribution model at BlaBlaCar

    Marketing attribution models are used to attribute conversions to marketing channels and hence monitor the performance of online marketing investments. Rules-based attribution models are frequently used in the industry with models such as ‘last interaction’, ‘first interaction’ or ‘position based’. They all follow a simple set of rules to perform the attribution, however the rules are set arbitrarily and may therefore be biased. At BlaBlaCar, we developed a data-driven Multi-Touch attribution model based on Markov chains. We will present the challenges that we have overcome from the idea to the roll-out and the impact on performance marketing of this project.

    Ken is a Data Analyst at BlaBlaCar with a master in Business Management and a Msc in Data Science from Edhec Business School. His business and data science background helps the understanding of business application of data science projects and smoothen the communication between technical data teams and business teams. He is working on various data-related marketing subjects such as the development of a Multi-Touch attribution model based on Markov Chains and the implementation of a Marketing Mix Modeling.

    Linkedin
  • 15:05

    COFFEE & NETWORKING BREAK

  • 15:45
    Ben Auffarth

    Experimentation at Scale at Loveholidays

    Ben Auffarth - Head of Data Science - Loveholidays

    Down arrow blue

    Experimentation at Scale at loveholidays

    loveholidays is the UK’s fastest-growing online travel agent. A core part of our culture of testing and learning is our in-house experimentation platform, Octopus, that has enabled us to run tests at no external cost with complete ownership of the data. In data science, we are running the automated analysis of this growing programme of a/b testing with dozens of concurrent experiments at any given time across different products, point of sales, and target groups. We work on reducing bias, and developing statistical procedures suited to measuring effects at low bias and reduced lag. In this document, we will go through our approach and the challenges we are facing.

    Ben Auffarth is the head of data science at loveholidays. With a background and Ph.D. in computational and cognitive neuroscience, he has designed and conducted wet lab experiments on cell cultures, analysed experiments with terabytes of data, run brain models on IBM supercomputers with up to 64k cores. More recently, he's built production systems processing hundreds of thousands of transactions per day, and trained neural networks on millions of text documents. He's authored two books on machine learning. When he's not at work, you might find him on a playground with his young son in West London. He co-founded and is the former president of Data Science Speakers, London.

    Linkedin
  • CUSTOMER SERVICE

  • 16:10
    Merve Alanyali

    Input Data Quality control with LV=’s Input-Checker

    Merve Alanyali - Head of Data Science Academic Partnerships and Research - LV= General Insurance

    Down arrow blue

    Input data quality control with LV=’s Input-Checker

    During the production of machine learning models, the quality of the data being received is crucial to return precise and trustworthy predictions. Without thorough checks in place, there is no end to the number of erroneous data points that could be passed by the system that the models are deployed to. But how can our model sniff out these issues? LV=’s input-checker offers an answer to these questions. In my talk, I will be introducing our python package input-checker; a light weight open source tool providing data checks at the point of inference.

    Dr Merve Alanyali is Head of Data Science Academic Partnerships and Research at LV= General Insurance drawing on an interdisciplinary background in computer science, complex systems and behavioural science. Merve has a PhD in Data Science from University of Warwick, fully-funded by the Chancellor’s International Scholarship. Her research at The Alan Turing Institute and University of Warwick focuses on analysing large open data sources with the concepts from image analysis to machine learning to understand and predict human behaviour at a global scale. The examples include identifying protest outbreaks using Flickr pictures, estimating household income with Instagram pictures and predicting non-emergency incidents in New York City. Her work has received more than 100 citations and featured on television and press worldwide including coverage in the Financial Times and Bloomberg Business. Prior to her PhD, she was awarded a double Masters degree in Complex Systems Science by the University of Warwick, UK and Chalmers University of Technology, Sweden.

    Twitter Linkedin
  • 16:35

    Open Q&A with Some of Todays Speakers!

  • 17:00

    NETWORKING RECEPTION

  • 18:00

    END OF DAY 1

  • 08:00

    DOORS OPEN & LIGHT BREAKFAST

  • 09:00
    Dino Bernicchi

    WELCOME NOTE

    Dino Bernicchi - Data Science Strategy Consultant -

    Down arrow blue

    Using Computer Vision with Recommender Systems to Transform Retail

    There has been an explosion in the number of e-commerce stores, available products, and well-crafted content. Recommender Systems have therefore come to the forefront in attracting customers, matching them to products they will love, and to content they will find engaging. But what happens when you don’t have many data points on your customers or products – or worse – customers are browsing your online store anonymously. How do you still deliver an engaging experience that drives your bottom line? Join this talk to discover how Computer Vision, Natural Language Processing, and other AI technology is being used with Recommender Systems to transform retail.

    Dino Bernicchi has 12 years’ experience driving AI strategy and developing productionized data solutions for large South African corporates such as TFG, Woolworths, Standard Bank, and Chevron SA. Most recently, as Head of Machine Learning at Homechoice - a home-shopping and financial services group - a successful AI strategy has driven R140m in additional operating profit at 15x return on investment. Dino is a regular speaker at AI, and data conferences, and lectures the Data Science Leadership for Executives course at University of Cape Town's Graduate School of Business. He also consulted on the development of course content for Andrew Ng's DeepLearning.AI.

    Linkedin
  • PERSONALISATION & RECOMMENDATIONS

  • 09:10
    Sharon Gieske

    Predicting Article Demand With Temporal Fusion Transformers

    Sharon Gieske - Data Science Tech Lead - Picnic Technologies

    Down arrow blue

    Predicting article demand with Temporal Fusion Transformers

    Picnic is the world's fastest growing online supermarket that makes grocery shopping simple, fun, and affordable for everyone. To ensure the freshest products and reduce waste, Picnic operates as a just-in-time supply chain. This must be balanced against high availability requirements for grocery items, as one unavailable product might lead to the loss of an entire basket. Accurate article demand forecasts are paramount. In this talk, we'll share how Picnic optimizes article demand forecasts with ML models. We'll dive deeper into why we transitioned from tree-based models to deploy the more state-of-the-art Temporal Fusion Transformer model and how it's used to balance waste & availability.

    Sharon Gieske is the Tech Lead of the Data Science team at Picnic. Here, she leads the development of technologies that enable Machine Learning projects in a wide range of domains such as demand forecasting, personalization, finance and more. Together with her team of data scientist, she works full-stack and builds end-to-end ML solutions to take e-commerce at Picnic to the next level.

    Linkedin
  • 09:35
    Calvin  Seward

    Applying Reinforcement Learning to Zalando's Recommender System

    Calvin Seward - Research Scientist - Zalando

    Down arrow blue

    Why you're Already Doing Reinforcement Learning and a Few Tips on Doing it Better

    “We’d like to do reinforcement learning, but...” I hear it all the time from ML practitioners working in fields as diverse as search, pricing, real-time bidding and recommendations. The funny thing is by creating search results, prices and recommendations these practitioners are already “doing” reinforcement learning (RL). They are creating policies which react to observations and collect as much reward as possible. Since they’re already doing RL, the question is how they can do it better. Those are the two points I’ll cover in my talk. First I’ll work though real world examples showing RL is both a set of problems that data leaders commonly encounter and a set of solutions tailored to solve those problems. Then I’ll dive into one specific and easy way to solve RL problem: the contextual bandit. When my session is done, you’ll not only know that you’re doing RL, you’ll also know how to do it better.

    Calvin Seward is a Research Scientist for Zalando's Research Lab, working mainly on reinforcement learning problems. By treating problems as diverce as pricing, recommendations and search as reinforcement learning problems, solutions can continually learn more about our customers and our products. The main line of attack for these problems is deep neural networks written in the Torch framework and trained on GPU clusters. At the same time, he is involved in applying lessons from GPU-driven HPC computing and cutting edge Machine Learning to other fields of the Zalando universe. Past projects have included sizing solutions for online retail, warehouse management and recommendation systems.

    Linkedin
  • 10:00
    Sam Drury

    I Love Wotsits!: How AI micro-decisions Are Creating Consumer Smiles at PepsiCo

    Sam Drury - Advanced Analytics Senior Manager - PepsiCo

    Down arrow blue

    I Love Wotsits!: How AI micro-decisions Are Creating Consumer Smiles at PepsiCo

    You might think it’s lucky that your local store has that bag of Wotsits and bottle of 7up that you enjoy every lunch time. In fact, AI tools have guided them every step of the way to put that smile on your face! Learn how PepsiCo use an advanced in-house data foundation to deploy image recognition, assortment optimisation, digital selling, and other AI tools that drive micro-decisions throughout the selling process.

    Sam Drury is Senior Manager for Commercial Advanced Analytics at PepsiCo Europe. He specialises in solving complex business problems with data. Sam manages teams in the development, deployment, and scaling of machine learning and AI solutions across Europe. Projects include go-to-market automation, image recognition, assortment optimisation, asset allocation, visit optimisation, acquisition targeting, and customer segmentation. Sam has worked across the globe with companies large and small, spanning retail, aviation, logistics, transport, supply chain, payments & fintech, digital marketing, customer relationship management, public services, environmental, leisure and tourism, yield management, IT, web, and beyond.

    Linkedin
  • 10:25

    COFFEE & NETWORKING BREAK

  • AI APPLICATIONS IN INDUSTRY

  • 10:55
    Juwel Rana

    Price Personalization Might Be The Next Big Thing for Retailers

    Juwel Rana - Head Of Analytics - Varner

    Down arrow blue

    Price Personalization Might Be The Next Big Thing for Retailers

    With advanced AI, retail industries are pushing personalization to the next level. Now it is not only recommending the right product for the right customer at the right time, but also taking personalization to the right pricing level. Following the trend of my previous talk in Re-work on personalized product recommendation, in this talk, the focus will be on sharing some insights on pricing aspect and how AI can be useful for personalized pricing.

    Juwel Rana, PhD is a global analytics leader located at Oslo, Norway and leads the analytical development at Varner Group, holding the position as Head of Analytics. He is AI product focused, believer of value driven development. Juwel possess full stack development principles from architecting AI product to algorithmic design as well as strategic and business objective setting. When it comes to AI technology, Juwel thinks business first, and when it comes to business, Juwel puts customer first.

    Twitter Linkedin
  • 11:25
    Celine Xu

    When Machine Learning Meets Fashion

    Celine Xu - Lead Data Scientist - H&M Group

    Down arrow blue

    When Machine Learning Meets Fashion

    Introduce the conflicts between Fashion and Machine learning; Summarise the limitation of current image recognition capabilities for fashion feature detection; Share ideas about how could industry better leverage machine learning in fashion industry.

    Celine is a lead data scientist from H&M group, who is passionate about leveraging machine learning and new technologies to support decisions, to improve company performance and to drive company digitalization. Celine has 9+ years’ experience in AI and advance analytics, focusing on fashion, retail, and banking industry. She has an applied mathematics degree from China and MBA from UK.

    Linkedin
  • 11:55
    Rick Rijkse

    Reducing Food Waste By Demand Forecasting at Scale

    Rick Rijkse - Principal Data Scientist - Ahold Delhaize

    Down arrow blue

    Reducing Food Waste by Demand Forecasting at Scale

    Ahold Delhaize is one of the world’s largest food retail groups. We use in-house built machine learning algorithms for demand forecasting, to reduce food waste and improve product availability. We make an extremely large number of forecasts every day, creating forecasts on SKU level for each store on a variety of forecasting windows. In this presentation, we will explain some of the challenges we dealt with when developing the models. We will share differences and similarities in approach for e-commerce vs. brick-and-mortar stores. Finally, we will discuss the solution design and the orchestration of all models that are in production.

    Rick Rijkse is a Principal Data Scientist at the Global Data & Analytics team of Ahold Delhaize. As expert on forecasting, he supports a variety of data science teams within the organization in implementing and improving forecasting algorithms. He has 10 years of experience in data science in different industries (advertising, finance and retail) and holds a master’s degree in Econometrics from the Erasmus University of Rotterdam.

    Linkedin
  • 12:25

    LUNCH

  • 13:25
    Dino Bernicchi

    Using Computer Vision with Recommender Systems to Transform Retail

    Dino Bernicchi - Data Science Strategy Consultant -

    Down arrow blue

    Using Computer Vision with Recommender Systems to Transform Retail

    There has been an explosion in the number of e-commerce stores, available products, and well-crafted content. Recommender Systems have therefore come to the forefront in attracting customers, matching them to products they will love, and to content they will find engaging. But what happens when you don’t have many data points on your customers or products – or worse – customers are browsing your online store anonymously. How do you still deliver an engaging experience that drives your bottom line? Join this talk to discover how Computer Vision, Natural Language Processing, and other AI technology is being used with Recommender Systems to transform retail.

    Dino Bernicchi has 12 years’ experience driving AI strategy and developing productionized data solutions for large South African corporates such as TFG, Woolworths, Standard Bank, and Chevron SA. Most recently, as Head of Machine Learning at Homechoice - a home-shopping and financial services group - a successful AI strategy has driven R140m in additional operating profit at 15x return on investment. Dino is a regular speaker at AI, and data conferences, and lectures the Data Science Leadership for Executives course at University of Cape Town's Graduate School of Business. He also consulted on the development of course content for Andrew Ng's DeepLearning.AI.

    Linkedin
  • 13:50
    Kelcey Jasen

    Predicting Monetary Lifetime Value of Customers

    Kelcey Jasen - Lead Data Scientist - MATCHESFASHION

    Down arrow blue

    Predicting monetary lifetime value of customers

    The customer lifetime value (customer LTV) project was prioritised to align with the MATCHESFASHION goal of driving marketing efficiency. Customer LTV is the predicted monetary spend per customer for the next 12 months. Customer LTV aids stakeholders with the data needed to support decisions to reach the right customers with the right messages. Our approach was to use the traditional method of modelling customer LTV which executes statistical models known as “Buy 'Til You Die” (BTYD) models. In our case of e-commerce, we do not observe a flag when a customer decides to stop shopping with us so we have to use tools that make assumptions on the declining probability of future purchases. The BTYD models use parametric distributions and strict assumptions to model customer LTV for customers who have made multiple purchases and provides the likelihood a customer will shop with us again in addition to, their predicted monetary spend over the next year. We will discuss our definition of customer LTV, use cases, model outputs, limitations, and roadblocks we encountered.

    Kelcey Jasen is a Lead Data Scientist at MATCHESFASHION – a global luxury fashion retailer. Kelcey’s primary focus is how to use data science to make customers at the heart of all our thinking. Previous to MATCHESFASHION, Kelcey was a Data Scientist at a PropTech start up in London and a Statistics instructor at a major university in the United States.

  • 14:15

    Panel: What Can Be Expected From The Future Of Retail

  • Rob McKendrick

    MODERATOR

    Rob McKendrick - (Former) Head of Data - Co-op

    Down arrow blue

    Bio:

    Rob is Head of Data at the Co-op, with end to end responsibility for every part of the data value chain from governance through engineering to analytics, insights and data science. The Co-op is one of the world's largest consumer co-operatives supporting members, customers and communities. Rob has worked in data and analytics for over 20 years, starting as a research engineer into using data to mine large databases for counter fraud and terrorism. Over the years he has worked at shaping and building data and analytic capabilities in household name organisations in Finance, Telecoms, Utilities, Government and Retail. The big attraction of the Co-op to Rob is that it is an open organisation owned by its members, which raises the bar on data governance and stewardship.

    Linkedin
  • Mark Bakker

    PANELIST

    Mark Bakker - Senior Account Executive - Sama

    Down arrow blue

    Mark Bakker is an xperienced and multi-lingual EMEA Account Executive with an Enterprise Solutions Manager/ Data Science background anda proven track record of exceeding performance metrics at both established corporate companies and startup environments. Mark is a program and change manager, creative, driving complete business development cycle through to close.

    Linkedin
  • Sharon Gieske

    PANELIST

    Sharon Gieske - Data Science Tech Lead - Picnic Technologies

    Down arrow blue

    Predicting article demand with Temporal Fusion Transformers

    Picnic is the world's fastest growing online supermarket that makes grocery shopping simple, fun, and affordable for everyone. To ensure the freshest products and reduce waste, Picnic operates as a just-in-time supply chain. This must be balanced against high availability requirements for grocery items, as one unavailable product might lead to the loss of an entire basket. Accurate article demand forecasts are paramount. In this talk, we'll share how Picnic optimizes article demand forecasts with ML models. We'll dive deeper into why we transitioned from tree-based models to deploy the more state-of-the-art Temporal Fusion Transformer model and how it's used to balance waste & availability.

    Sharon Gieske is the Tech Lead of the Data Science team at Picnic. Here, she leads the development of technologies that enable Machine Learning projects in a wide range of domains such as demand forecasting, personalization, finance and more. Together with her team of data scientist, she works full-stack and builds end-to-end ML solutions to take e-commerce at Picnic to the next level.

    Linkedin
  • 15:00

    CLOSING COFFEE BREAK

  • 15:30

    END OF SUMMIT

This website uses cookies to ensure you get the best experience. Learn more