Hybrid Image Search at Scale: Lessons in Accuracy, Latency, and Cost

When users can’t describe what they want, they show it. Image search has emerged as a powerful way to capture user intent in e-commerce. But building a system that is accurate, fast, merch oriented and cost-effective at industrial scale is no easy feat.

In this talk, we’ll share how the Search & Publication team at Adeo (Leroy Merlin group) built a scalable hybrid image search engine serving over 10 million products, combining visual embeddings, textual signals, and knowledge-graph-enhanced metadata.

We will share with you:

• How we built a hybrid architecture combining image embeddings and LLM based lexical search, backed by our Knowledge Graph.

• Our journey into vector quantization techniques (bf16, int8, int4, int2, BBQ in Elasticsearch), and how they impacted latency, precision and cost trade-offs.

François Gaillard

Adeo Services

Engineering Manager at Adeo (Leroy Merlin, Bricoman) in AAAI team (Advanced Analysis and Artificial Intelligence) in Search & Publication domain. Last 10 years building and running AI powered search engines for e-commerce websites. PhD in computer science, specialized in Artificial Intelligence

Guilherme de Freitas Guitte

Adeo Services

Engineering Manager at ADEO (Leroy Merlin, Bricoman), technical leader in recommendation systems. With 15 years of experience in developing publication systems for e-commerce platforms, he now drives and builds the technical vision / strategy of an omnichannel, multi-country recommendation engine. He leads a multidisciplinary team combining software development, data engineering, and data science, with a strong focus on cloud, scalability, and deploying ML models into production.