Interview #9 of our St. Patrick's Day series, and we're heading to Stockholm.
Jim Dowling is the co-founder and CEO of Hopsworks, the European answer to Databricks. A Dublin native who spent 11 years at Trinity College Dublin before moving to Sweden for love, Jim has built one of Europe's most important AI infrastructure companies from the heart of Scandinavia. Hopsworks helps global enterprises build and operate AI systems at scale, with customers ranging from Zalando to a Fortune 100 US bank.
From Trinity's computer labs to taking on Databricks and Palantir, this is some story.
Stay tuned for our final interview on Monday and keep an eye out for our St. Patrick's Day newsletter, where we reveal the full results of our survey with over 120 Irish founders across the globe.
Lay All Your Love On Me 🎼

Jim Dowling
How did you end up in Sweden, Jim?
It is often said that people emigrate for either money or love. I left Ireland in 2005 for Sweden, and It definitely wasn’t for the money. I spent most of the first thirty years of my life in my hometown of Dublin, with 11 years in Trinity College Dublin, starting from a BA in Computer Science (1st class honors), to a PhD, to a lecturing position. I had close proximity to many startups from Trinity from Iona Technologies to Havok to Intercom. My PhD supervisor Vinny Cahill was Iona’s first CTO and when I joined his research group, DSG, the Iona founders had all just left. Havok came from the neighbouring vision lab that left on mass to build a lasting platform for computer gaming. And as a lecturer I supervised Ciaran Lee for 12 months (on using machine learning to predict wave heights for surfing!) who went on to co-found Intercom.
Then life got in the way, and I moved to Sweden for my now wife. I got a job in Stockholm at a database company called MySQL, which at the time was a popular open-source database. MySQL was acquired in 2009 by Sun Microsystems, and although I had some shares I didn’t even make enough money to buy a car. The vast majority of the 200 employees had a similar experience, though the founders cashed out big time. Unsurprisingly, post-acquisition, the MySQL diaspora scattered just when data was exploding as a category. The nature and timing of that acquisition had an impact on me about how to build a founding team and how to value key talent when I co-founded Hopsworks in 2018.
How did Hopsworks come to be?
Hopsworks was the output of systems research work performed at KTH - Royal Institute of Technology, where I took a lecturing position in 2011, and the Research Institute of Sweden (called SICS at that time), where I was a senior researcher since 2007. I went back into research after being awarded a Marie Curie Intra European Scholarship.
Hopsworks was founded by myself, six of my students at RISE and KTH, and Prof Seif Haridi in 2018. We raised our initial venture capital round from a Finnish investor, Inventure, and Frontline Ventures (with Steve Collins, ex TCD and Havok CEO). Hopsworks is a platform for building and operating AI systems at scale. Its key differentiations are a scalable file system, HopsFS (IEEE Scale Prize in 2007), as well as the open-source MySQL Cluster database. We quickly forked the database and brought in its inventor, Mikael Ronström, and developed it as our own database RonDB.

Jim and his recently publihed book ‘Building Machine Learning Systems with a Feature Store: Batch, Real-Time and LLMs’.
The business and the future 🔮
What does Hopsworks actually do?
Hopsworks is used by global Enterprises to help them build and operate AI systems at scale. It is best known as the category creator of feature stores for machine learning. Feature stores are a data platform that manages and transforms data for training and inference in AI systems. A well known customer is Zalando, Europe's largest fashion e-retailer. They use Hopsworks to collect and transform the events that users create on the website and mobile applications. Hopsworks helps turn that data into training data for models and also feeds that data to machine learning models that do things like personalize recommendations for users.
Hopsworks provides software infrastructure support for the whole AI lifecycle - from data transformations (in Python, Spark, etc), to training on GPUs, to model inference (serving LLMs), to agents. The closest platforms to Hopsworks are US-based, such as Databricks, GCP Vertex, and Sagemaker. In 2024, we published a research paper at the leading database conference SIGMOD showing we outperformed all these platforms by a significant margin. Hopsworks is a Kubernetes native platform, and we can deploy anywhere. That makes Hopsworks a "sovereign" Data/AI platform, as it can be deployed on private data centers. We have customers such as a large law enforcement agency in Europe, and the largest defence manufacturer in the Nordics, Saab. Given the recent political calls for sovereign technology in the EU, Hopsworks could have an important role to play as the only scalable Data/AI platform in Europe.
What are your top priorities for the next 12 months?
Even though it is satisfying to have lighthouse customers in Europe, our largest customers are in the US, including a Fortune 100 bank. We are still a global company. In 2026, we plan to expand in both the US and Europe. Our main goal is to be profitable by the end of 2026. The days of chasing growth at any cost are behind us. Although the fundraising cycle is very hyped for AI companies, it is primarily centered on a few agentic applications and large language model developers. As an enterprise software infrastructure company, we haven’t been exposed to the consumer AI boom to the same extent as other companies from Stockholm, such as Lovable.
What is the biggest opportunity for you over the next 12 months?
We see massively increased demand in both our existing customers and prospects for a converged Data and AI platform. Hopsworks has been one of the few global companies that provide a unified platform to manage your data at scale for both AI and analytics. We recently extended Hopsworks to support analytics, adding dashboards and SQL query over Lakehouse data. Our aim is to reposition the product as the global leading sovereign Data/AI platform - an on-premises Databricks if you will. And yes, Palantir is a competitor there - a challenge we take on with relish. The upcoming Cloud and AI Development Act by the European Commission could have a large impact if it includes a mandate for EU public procurement to buy European. The EU AI Act will soon come fully into force and we are ahead of our (American) competitors on that front.
The Irish bit ☘️
Has being Irish helped your business?
Hopsworks is becoming more Irish, with Raymond Cunningham, our VP Customer Success, joining the board. Being Irish has definitely helped me build a European enterprise software company that sells in the US and across the world. Not just through language and culture, but in our diaspora who are spread across the world’s leading tech hubs and always help provide an introduction or advice that can, in the end, make all the difference.

Know a founder who deserves to be featured? Reply to this email. 🤝
Was this forwarded to you? Subscribe to Irish Insights and join the community.