Qdrant brings applied IA solutions to the next level making metric learning practical. Our flagship product – neural search engine – provides a production-ready service with a convenient API to store, search, and manage vectors along with the additional payload.
Qdrant is tailored to extended filtering support making it useful for all sorts of neural-network or semantic-based matching, recommendations, faceted search, and other applications.
Tell us about yourself?
I have over 15 years of experience in IT. I studied Computer Science in Karlsruhe, Germany. Started my career as a backend developer in Berlin, working on scaling services at one of the biggest social networks in western Europe, worked as a social layer developer for a big gaming company in Hamburg, worked for a VC helping startup founders to build tech products. Before founding Qdrant I spent 5 years leading the tech and product team as CTO/CPO at a startup building recruiting as a service solution with an applied AI solution under the hood that matches candidates with jobs. The job matching requires a combination of unstructured data along with some hard filters. There were no tools to solve this challenge, so we had to build sub-optimal workarounds. Later we recognized that there are a lot of similar challenges in different industries. This is how the idea of Qdrant, a neural search engine with extended filtering support, was born.
What is the inspiration behind your business?
The traditional, classification-based Data Science approach has a lot of disadvantages like cold start issues with just a little amount of data, expensive and time-consuming manual labeling, need for continuous retraining, etc. As a consequence, it leads to high maintenance complexity, high resource demand, and high costs. The solution is the Deep Metric Learning approach with the following advantages: Self-supervised, no manual labeling required, on-the-fly new data entries without retraining, easily tweaking. With this approach, based on vector similarity, building applied IA solutions becomes easier, more flexible, and cost-effective.
What is your magic sauce?
Qdrant is not the only semantic similarity search engine out there. There are already some solutions on the market. Some of them are also open-source, some not. However, Qdrant is the only solution with extended filtering support, unbeatable performance, the highest scalability offering, and the lowest integration barrier at the same time. These factors make it possible for Qdrant to become the global standard for neural similarity-based AI applications.
Where do you see your company going in 5 years time?
The vision is Qdrant to become the de-facto global standard for neural similarity, matching, recommender solutions, and semantic search and provide a scalable and easy-to-use service for businesses to take advantage of AI and make the most of their unstructured data.
What has been your biggest setback so far?
Building a deep tech company in Europe is not easy, especially if you are following the open-source approach. Most of the investors do not understand it or are not really interested in technical innovation and long-term visions and prefer to invest in short-term hyping trends. We really hope it is going to change in the future and we want to be a part of this movement.
What is the next big challenge for your business?
We are at the very beginning of our journey and just started to build the company, looking for skilled, enthusiastic colleagues to join our team. Our main focus at the moment is on growing the community and gaining trust and early adoption in parallel with building out our technical superiority. Also, we are looking for partners in the industry and are open to cooperation.
How do people get involved/buy into your vision?
We are looking for early adopters and welcome all interested in Data Science and Machine Learning to join our community. Follow Qdrant on GitHub https://github.com/qdrant/qdrant, check out the documentation on Qdrant’s homepage https://qdrant.tech. Also, if you are into Rust take a look at the Qdrant engine code base and feel free to give us some feedback or even contribute to our development.