Technical Talks


AI Launchpad 2025: TopK
Missing value detected...
Video will be populated after the conference
TopK's platform drastically simplifies the modern search stack into a single API. The team spent three years building one of the market's most popular vector databases, realizing that building a database around vectors was the wrong abstraction and that production-ready search extends far beyond retrieval. This insight led to the creation of TopK, beginning with a new search database that unifies dense/sparse vector + text + faceted retrieval with custom scoring in a single query. By building both the query execution engine and storage format from scratch, TopK gained significant flexibility and cloud cost-efficiency. Their next step involves integrating higher levels of the search stack into the API to drastically reduce time-to-value for customers.

Co-Founder
Marek Galovic
TopK
Marek likes to break assumptions and push the boundaries of what's possible when building from first principles. Before founding TopK, Marek spent three years at Pinecone where he led data plane and control plane engineering. Prior to that, he worked on fraud detection and forecasting algorithms at Shopify and researched applications of game theory to make malware classifiers more robust at CTU/Avast.

Co-Founder
Jerguš Lejko
TopK
Jerguš cares about developer experience, fast feedback loops, and efficient delivery. At Pinecone, he built developer tooling, optimized CI/CD for snappy builds, and managed cloud infrastructure across all major hyperscalers. Before that, he worked with ML and bioinformatics teams, building data pipelines and automation for large-scale processing. Now, Jerguš is building TopK, a data infrastructure company to radically simplify search and make it more accessible to developers.
Discover the data foundations powering today's AI breakthroughs. Join leading minds as we explore both cutting-edge AI and the infrastructure behind it. Reserve your spot at before tickets sell out!