Bavaria, Advancements in SEarch Development (BASED) Meetup
About us
Search is evolving fast. From new algorithms and tools to AI-driven solutions, the field is constantly shifting.
BASED Meetup is where professionals & enthusiasts discuss the latest trends, breakthroughs, and challenges in modern search.
We dive into topics like:
- Agentic Retrieval-Augmented Generation (RAG), Memory & Search;
- Innovative approaches to retrieval;
- Modern metrics for search quality;
- Optimization techniques for production-scale systems;
- Cross-lingual and multimodal search challenges;
… and much more.
No sales pitches, no gatekeeping — just an open space to share ideas, learn from each other, and explore the technologies shaping the future of search!:)
P.S. We now have #BASED stickers!
Agenda
18:00 — Doors Open
18:15 - 18:25 — Welcome Note (Daniel Wrigley, Evgeniya Sukhodolskaya, TNG Technology Consulting)
18:25 - 18:55 — "Image-Based Product Recommender Using Vector and Attribute Search" (Florian Janoschek, TNG Technology Consulting)
18:55 - 19:00 — Break
19:00 - 19:30 — "Reinventing Relevance Feedback: Scalable, Cheap, Universal" (Evgeniya Sukhodolskaya, Qdrant)
19:30 - 20:00 — "Agentic Relevance Tuning: Faster, Easier, and Higher Quality Search at Scale" (Daniel Wrigley, OpenSource Connections)
20:00 - 20:01 — Closing Note (Daniel Wrigley, Evgeniya Sukhodolskaya)
20:01 - 21:00 — Networking with drinks & pizza sponsored by TNG Technology Consulting
Talk Descriptions:
Talk 1: Image-Based Product Recommender Using Vector and Attribute Search
Speaker: Florian Janoschek (TNG Technology Consulting)
Abstract: For a fashion retailer, we developed a system that suggests alternative and complementary products based on product images. The solution leverages Elasticsearch and its k-nearest neighbour search
capabilities. In this talk, I will share practical insights and
demonstrate how to integrate vector search with attribute-based sorting and filtering.
Talk 2: Reinventing Relevance Feedback: Scalable, Cheap, Universal
Speaker: Evgeniya Sukhodolskaya (Qdrant)
Abstract: The relevance of retrieved results is quite a metric to deal with, results returned to a request could be meh and more meh... And not even the best agent can help it.
Improvements often lie in the use of IR tools such as the relevance feedback approach, well known in lexical search.
This approach could also benefit dense retrieval immensely. But how? In production, dense retrieval on heavy encoders alone is unfeasible: too slow & expensive. Imagine that a simpler, faster retriever could be guided by a smarter model (from a cross-encoder to a full-blown agent) during vector index traversal.
I’ll talk about this method of incorporating relevance feedback into production systems, so it’s built directly into the vector search engine.
Talk 3: Agentic Relevance Tuning: Faster, Easier, and Higher Quality Search at Scale
Speaker: Daniel Wrigley (OpenSource Connections)
Abstract: Search relevance tuning is one of the hardest—and most expensive—problems in modern search systems. It demands deep expertise in the search engine's scoring, complex query DSLs, and iterative manual testing. While there are tools and techniques available that provide the data and the environment for improvement, the leap from "analyzing data" to "deploying a fix" remains a significant hurdle for many search engineers.
Agentic Relevance Tuning (ART) is an end-to-end framework that uses LLM-powered agents to automate the full relevance lifecycle. ART replaces time consuming manual workflows with a coordinated system of specialized agents that continuously detect relevance issues, hypothesize improvements, and orchestrate offline and online evaluation.
By combining relevance tooling with the reasoning and orchestration capabilities of agents, ART transforms relevance tuning into a faster, more accurate, and more accessible process—enabling both beginner and expert search engineers to ship better search experiences to their customers.
Venue
We're hosted this time by TNG Technology Consulting, who also sponsor networking drinks and snacks.