Kirill Shvidler Harnesses the Power of GenAI for Artera's Art Discovery Platform
Kirill Shvidler, Data Scientist at Artera
Data Scientist at Artera
Kirill Shvidler is the Data Scientist and Solutions Architect at Artera, being one of the first 3 founding members Kirill was responsible for building the core cloud infrastructure on AWS and Astra as well as developing most of the machine learning features used in the Artera app, including Image Capture, Hybrid Semantic Search and Deep Artwork Recommendations.
Transcript
Artera is a social platform where you can share, explore and discover art. You can interact with art using multimedia comments, using audio or audio text descriptions.
We are constantly experimenting with new tech and new cool stuff which can make Artera fun to use, but also in a way that can be helpful and meaningful for artists.
So we are dedicated to supporting them and making their life better. We are using deep-learning embeddings for our artworks so that you can explore similar artwork recommendations and see content that you might enjoy. We're also using AI in our image capture technology, which allows you to take a snap or a photo of a real-life object or a painting and find similar artworks or find an exact match.
Some of our most exciting features on Artera are built on vector search. Throughout the whole app, we're using Astra DB to give you similar items like artwork or comments, and basically make your life easier as a user and make it easier for you to find what you want. Astra DB is a serverless product, and we have a very small team we would much rather focus on solving actual business problems instead of worrying about our infrastructure all the time. Astra DB makes it very easy to scale without having to worry about any of the infrastructure, which makes our whole team's life much, much easier.
We considered quite a few other alternatives. We considered vector databases like Pinecone, Zilliz, and OpenSearch. However, we were always building on top of Cassandra, we were using Astra DB even before they had vector search capabilities. With the introduction of Astra DB, we managed to scale our systems and improve some accuracy on existing technologies like image capture. It's also made the developer experience much better.
We're developing a lot of new features, for example, next-generation patronage, peer to peer art education, and most recently, we are developing a unique representation of your own art taste that's built on Astra vector search. I personally believe is that data quality is always more important than data quantity. The better your data is, the better your product is. That's why maybe streaming solutions can provide you an edge over others. And in general, having better data means leading in this industry.
Artera is a social platform where you can share, explore and discover art. You can interact with art using multimedia comments, using audio or audio text descriptions.
We are constantly experimenting with new tech and new cool stuff which can make Artera fun to use, but also in a way that can be helpful and meaningful for artists.
So we are dedicated to supporting them and making their life better. We are using deep-learning embeddings for our artworks so that you can explore similar artwork recommendations and see content that you might enjoy. We're also using AI in our image capture technology, which allows you to take a snap or a photo of a real-life object or a painting and find similar artworks or find an exact match.
Some of our most exciting features on Artera are built on vector search. Throughout the whole app, we're using Astra DB to give you similar items like artwork or comments, and basically make your life easier as a user and make it easier for you to find what you want. Astra DB is a serverless product, and we have a very small team we would much rather focus on solving actual business problems instead of worrying about our infrastructure all the time. Astra DB makes it very easy to scale without having to worry about any of the infrastructure, which makes our whole team's life much, much easier.
We considered quite a few other alternatives. We considered vector databases like Pinecone, Zilliz, and OpenSearch. However, we were always building on top of Cassandra, we were using Astra DB even before they had vector search capabilities. With the introduction of Astra DB, we managed to scale our systems and improve some accuracy on existing technologies like image capture. It's also made the developer experience much better.
We're developing a lot of new features, for example, next-generation patronage, peer to peer art education, and most recently, we are developing a unique representation of your own art taste that's built on Astra vector search. I personally believe is that data quality is always more important than data quantity. The better your data is, the better your product is. That's why maybe streaming solutions can provide you an edge over others. And in general, having better data means leading in this industry.