

Unstructured Data Meetup London
This Unstructured Data Meetup is timed to follow AWS Summit London (April 22), capturing momentum from conference attendees. On April 23rd, we bring together developers and technical decision-makers who are actively building AI applications with unstructured data, LLM, and agents — for an evening of deep technical exchange and relationship building.
🗓️ When: April 23, 2026 @5:30 - 8:30 PM
📍 Where: 15F, Amazon Office LHR16, 1 Principal Place, London EC2A 2FA
🤖 Who: This meetup will be catered to makers who are actively building AI applications with unstructured data, LLM, and agents.
💻 What to bring: An ID matching the name on your registration for entry.
Agenda
17:30 – 18:00 | Welcome & Networking (sign-in, drinks, open mingling)
18:00 – 18:05 | Opening Remarks (Zilliz Host)
18:05 – 18:35 | Talk 1: Building Agent Memory at scale with Milvus— Jiang Chen, Head of Developer Relations, Zilliz
18:35 – 19:05 | Talk 2: Conversational E-commerce: Recommender Systems in the Age of Shopping Assistants — Firas Jarboui, Head of ML, Gorgias
19:05 – 19:15 | Break & Networking
19:15 – 19:45 | Talk 3: [TBD]Hou Ying, UKI GenAI Specialist SA, AWS
19:45 – 20:15 | Talk 4: [TBD]
20:15 – 20:30 | Networking & Closing
Speakers & Session Details:
Jiang Chen, Head of Developer Relations, Zilliz
▶Presentation Title: Building Agent Memory at Scale with Milvus
▶Presentation Abstract: As AI agents move into production, the real challenge becomes building systems that can remember, retrieve, and scale. This talk explores how vector databases enable persistent agent memory beyond the context window — and what it takes to make it work under real-world constraints like latency, cost, and multi-tenancy.
Featuring practical patterns, production insights, and a walkthrough of the open-source memsearch project.
Firas Jarboui, Head of Machine Learning - Senior Engineering Manager, Gorgias
▶Presentation Title: Conversational E-commerce: Recommender Systems in the Age of Shopping Assistants
▶Presentation Abstract: This talk demonstrates how we built a product retrieval system that allows an AI shopping assistant to make better real-time decisions. We will explain how embeddings, re-ranking, and decision policies work together to let the assistant either confidently recommend products or ask clarifying questions when needed. The session also covers the use of real-world browsing behavior as a training signal to improve relevance across diverse merchant catalogs. Finally, it will highlight the implementation of confidence and uncertainty mechanisms that make the assistant more reliable, adaptive, and effective in production.
[TBD]Hou Ying, UKI GenAI Specialist SA, AWS
[TBD]
🚀 We look forward to seeing you at the meetup and talk about search!
Note: This event is sponsored by Zilliz (maintainers of Milvus). Special thanks to AWS for the venue and support, check out their upcoming event! By registering, you are agreeing that you will be respectful of event staff and act appropriately.
For more details: https://milvus.io/ja