Seldon was founded to help accelerate the adoption of machine learning to solve some of the world’s most challenging problems.
Since its launch six years ago, Seldon has raised €3 million in seed funding and been deployed in Google, RedHat and IBM machine learning platforms.
Machine learning has the potential to tackle a range of business problems, and Seldon’s product portfolio – Core, Deploy, and Alibi – are designed to give business owners the ability to seamlessly implement it.
Founder: Alex Housley
We heard from Alex about the story behind Seldon.
Why did you start Seldon?
Our aim is to accelerate the adoption of machine learning to solve some of the world’s most challenging problems, helping organisations transform their business and truly harness the power of their machine learning (ML) models to power their growth.
Prior to founding Seldon in 2014, we were building recommendation engines for media organisations and realised quickly that machine learning had the potential to bring huge growth to these businesses.
However, we were witnessing first-hand the challenges of scaling model deployment in these organisations.
That’s why we decided to take the company in a new direction and focus on creating the underlying technology that would allow these companies to run models reliably and efficiently at scale.
Since then, machine learning has been seen as increasingly important in solving business problems.
Data science teams need the freedom to work in any language or framework, but it’s equally essential for organisations to retain oversight and compliance across business functions.
It was important for us to establish credibility in the open source community where the strength of our technology could truly be tested in some of the world’s leading organisations before creating an enterprise-grade product that could truly transform how machine learning is deployed and trusted across the business community.
Tell us more about the tech behind the product.
Seldon’s technology allows businesses to deploy machine learning models into production faster, optimize the performance of ML models running in production and have better governance to reduce risk and reputational damage.
Machine learning has the capacity to transform an organisation by solving critical business problems.
However, while most organisations have reached the point where they understand this technology can bring huge benefits, they often falter in being able to reliably and efficiently put the technology into action in their organisations.
Compliance and risk management are often barriers to the success of this technology which can be easily tackled with Seldon.
That’s why we developed Alibi, an algorithm library for model explanations and monitoring, which delivers clear, unbiased insights into an organisation’s machine learning “black box”.
This means business executives with minimal coding expertise can understand the decision-making process of models handling sensitive data or making important customer-facing decisions.
Seldon offers a cloud native solution which brings together different models built in different environments and allows the data scientists in an organisation to work in any language or framework.
This allows organisations to easily scale their model deployments and improve collaboration with the DevOps engineers and the data scientists often tasked with having to interpret the complex models they create.
We just released an enterprise product, Seldon Deploy, which enables machine learning teams to move into production 5 times faster and more easily with compliance, audit trails, approvals, advanced experiments, and model explanations.
Built on the advanced ‘Seldon Core’ technology trusted by thousands of companies in the open source community, it is equipped with all the monitoring, user interfaces and compliance checks necessary for large scale organisations to deploy their models at scale and with confidence.
Where are you at right now?
We’re going through an exciting period of expansion.
We secured €3 million in a seed funding round led by Amadeus Capital Partners, which is being invested in Seldon’s enterprise product.
Since 2018, we have seen an incredible 25% month-on-month growth and over two million installs of our Seldon Core product to-date.
We’ve got a great team of internal engineers and community contributors behind the product, who are constantly working to develop the technology and ensure the Seldon platform delivers tangible benefits for all of our users.
Our client base is constantly growing, working with customers such as IBM, Google and Oracle on their machine learning platforms. Seldon also has an ongoing partnership with open source company RedHat.
What are your aims for the next year?
We have a number of exciting commercial deals that recently closed and are in the pipeline with some of the world’s biggest companies across sectors; so our team of over 20, all based in our London office, will at least be doubling and we’re planning to expand our team in the US.
What’s been the hardest thing about getting Seldon off the ground?
When we started creating Seldon, there was very little appreciation in the UK market for open source technology.
This is a model that has since been proven in the US, but at the time there was skepticism around the value of a ‘free’ product.
In the years since, we have seen business executives place increasing trust and, sometimes more importantly, budgets in the hands of their data science and technical teams which has allowed open-source products to make their way and prove their value in the world’s largest organisations.
The ecosystem has evolved considerably over the past few years and now we see tech leaders developing and sharing open source products.
We’re proud of our open source heritage, and continue to enjoy the collaboration and integrity of open-source projects like Kubeflow, which we’ve worked on with giants like Google and IBM.
We’re leveraging this shift to democratise technologies that were once only accessible to tech giants, and making them available to everyone.
That’s why Seldon was created to be framework-agnostic to support the use of agile workflows, so organisations aren’t forced into being locked in with one provider after only making a few initial steps in their roadmap.
Why should more people be using Seldon?
Our technology allows businesses to confidently deploy machine learning models in the most efficient way.
Across industry verticals, businesses can tackle difficult and complex problems at speeds previously unthinkable.
This is also all achieved with even more oversight and understanding throughout an organisation which means minimal risk in deploying faulty decisionmaking or relying on models built on bad data.
How much will it cost users – and why is it worth the investment?
We’ve proven time savings of up to 85% which can deliver a huge return on investment for businesses.
With a recent Fortune 100 client, we projected savings of around $15 million – and this doesn’t include the value of moving ML models into production 5 times faster.
We make it easy to get started with free open-source solutions that add real value, and price for our enterprise products and services based on usage and level of support required.
Our team of machine learning engineers help decrease the time-to-value and clearly measure the success from day one.