DDS

Case study · Data Engineering & AI

AI-augmented analytics at document scale.

Millions of documents scattered across dozens of indices, no unified schema catalog, no way to ask for a metric in natural language and get a reliable answer. I built the missing layer.

Updated · by Davide De Sanctis

01Approach

Collection and filtering pipeline, clustering, cohort analysis, AI-driven reprocessing.

Stack: Elasticsearch/Kibana, Qdrant for the vector/RAG layer, Redis.

02Result

Over 2 million total documents, including 1.5M+ activities and 350K users · 60+ real-time indices · 2,000+ fields mapped in the schema catalog · nearly a decade of historical data · 11 analysis dimensions.

Rounded figures: the project is under NDA.

Related

RAG chatbots on company documentsAssistants that answer from manuals, contracts, procedures and internal data, citing sources.LLM developerLLM-based products and features, with measurable quality and code you own.

Let's talk about your project

A free call to understand whether AI is the right answer, how big the project is and where to start.

Get in touch →