Elastic FR User Group

🎤 Want to submit a talk for an upcoming event? Feel free to come and share your ideas at https://ela.st/cfp-elasticfr. 🙏🏼

### Agenda

• Doors open
• Welcome words by Elastic and Amadeus (5 minutes)
• Talk 1: Test your Transistor: Visual Defect Detection by Alessandro Brofferio (20 minutes)
• Talk 2: Migrating an In-House Patent Search to Elasticsearch (french) by David Ventura and Laurent Quintian (20 minutes)
• Food & drinks

### Speakers & Talks

Test your Transistor: Visual Defect Detection with Elasticsearch and Vector Search
By Alessandro Brofferio, Elastic

Every transistor tells a story — but some stories end in a defect.

In this talk we'll show how Elasticsearch, vector embeddings, and a pinch of nearest-neighbour magic can automatically tell the good ones from the bad ones.

We'll demo a fully working KO/OK detection pipeline: Relying on reference images and let Elasticsearch's Inference API and KNN search do the heavy lifting. No model training, no GPU cluster — just smart search.

Come for the transistors, stay for the vectors.

Migrating an In-House Patent Search to Elasticsearch
By David Ventura and Laurent Quintian, Questel

This talk shares real-world lessons from migrating a large-scale patent search platform from a legacy proprietary engine to Elasticsearch. Patent search brings specific challenges: massive and highly structured documents, multilingual content, critical recall requirements, and regular high-volume data updates.
The session focuses on the engineering trade-offs behind relevance, scalability, indexing freshness, and production performance.