Accelerating AI innovation at Ahold Delhaize’s 2025 machine learning operations conference

MLOps

April 15, 2025 – This March, Ahold Delhaize hosted its second annual machine learning operations (MLOps) conference at the company’s head office in Zaandam, the Netherlands, bringing together over 130 tech professionals for a day of knowledge exchange and collaboration.

Part of Ahold Delhaize’s Growing Together strategy involves leveraging technology and data to drive customer innovation and operational excellence. With the growing popularity of AI, machine learning operations (MLOps) is developing rapidly as a field, and is critical for a food retail company like Ahold Delhaize and its brands – as well as many other tech-driven companies.

The conference provided an ideal platform for professionals to share insights and solutions while exploring the latest advancements in this rapidly growing sector. Focused on machine learning, the event highlighted the importance of optimizing AI model deployment and accelerating innovation across various industries by reducing time to production.

Key highlights from the speakers

This year’s event brought together inspiring speakers from some of the Netherlands' leading companies, including ABN AMRO, Adyen, Ahold Delhaize, Albert Heijn, bol, Booking.com and Eneco . These experts presented cutting-edge MLOps strategies and shared the challenges they are tackling in their respective fields. The conference covered three key topics that are crucial for the future of MLOps:

How infrastructure for MLOps should be organized
How to organize teams to speed up time to market for machine learning products
How to go about challenging machine learning projects

Building a GenAI platform at ABN AMRO
Rafael Pierre, GenAI Engineering Lead & Architect at ABN AMRO, introduced the company’s secure platform for deploying open-source large language models (LLMs). He explained the strategies to mitigate risks, including vulnerability scanning and centralized model management.

MLOps at Adyen
Asia Krajewska, MLOps Tech Lead at Adyen, shared the company’s approach to building reliable machine learning systems for payment processing, ensuring high availability and low latency. She explained how Adyen uses a mix of open-source and custom tooling to meet the demands of real-time payment transactions.

Scalable and secure data platform at Ahold Delhaize
Aoife O'Callaghan, Manager Data platform at Ahold Delhaize, showcased how the company is building a scalable and secure data platform using Azure and Infrastructure as Code (IaC). Key highlights included the integration of FinOps and fostering a culture of collaboration and self-service with robust security and scalability guardrails.

Workflow orchestration at Albert Heijn
Niels van den Berg, Senior Machine Learning Engineer at Albert Heijn, shared the team’s journey in automating workflow orchestration with Databricks. The session highlighted challenges faced and how Databricks Asset Bundles improved scalability in their operations.

Real-time product recommendations at bol
Binyam Gebre, Senior Data Scientist at bol, shared insights into the challenges of generating real-time recommendations for over 13 million customers and 40 million products. He detailed how bol uses Vertex AI and BigTable to maintain up-to-date user embeddings and ensure fast, accurate recommendations.

Ultra-low-latency feature serving at Booking.com
Başak Eskili, Machine Learning Engineer at Booking.com, presented the success story of their custom feature store, built to handle 200K RPS with latency under 25 ms. She explained the technical implementation and cost-efficient strategies that support real-time applications at scale.

Building MLOps teams in non-tech companies
Raphaël Hoogvliets, MLOps Tech Lead at Eneco, discussed strategies for establishing MLOps teams in non-tech environments. He emphasized leadership buy-in, long-term product thinking, and practical staffing strategies to ensure the long-term success of MLOps initiatives.

Deploying recommendation models with mono-repo workflows
Victor Pion, Data Scientist, and Jonathan Torres, Machine Learning Engineer, at Ahold Delhaize, discussed the company’s shift to a mono-repo workflow for deploying recommendation models. They showcased how this approach improved flexibility, simplified deployments, and accelerated iteration cycles for their teams.

LLM security and adversarial prompts
Folkert Stijnman, ML engineer at NS, provided an eye-opening talk on the growing threats of prompt injections and adversarial attacks on large language models. He emphasized the importance of proactive testing and developing defense mechanisms to safeguard AI systems.

Future collaboration

This year’s MLOps conference not only showcased the innovative work being done by Ahold Delhaize and its industry peers but also reinforced the growing importance of collaboration across sectors to solve the challenges of deploying and scaling machine learning models. The success of this event marks another milestone in Ahold Delhaize’s journey to harness the power of technology to enhance operations, improve customer experiences, and create new opportunities for growth. With the continued evolution of MLOps, the company is excited about the future of AI and its transformative impact on the retail industry and beyond.