02Multimodal retrieval infrastructure
ZS Associates · enterprise scientific documentsScientific Document Intelligence
10,000+ concurrent users · section retrieval within 5 minutes
A document-to-answer pipeline for long scientific files. It extracts structure from dense multimodal documents and makes useful passages available to downstream agents through section-aware retrieval tools.
- Parse 1,000+ page scientific documents, including complex structure
- Retrieve at document or section level through MCP tools
- Serve 10,000+ concurrent users; full parse within about 5 minutes
HIGH LEVEL DESIGNSELECT A STEP TO EXPLORE
STEP 01 / 05Scientific files
1,000+ page · multimodal
RUNTIME / SYSTEM LAYER
Amazon EKSCelery workers + message queues10,000+ concurrent users
From the constraint
to the system.
Built the document and section retriever tools over Azure Document Intelligence and custom XML handling; instrumented the workflow with Langfuse and deployed the retrieval workers on EKS with Celery and message queues.
END-TO-END PATH
01Multimodal scientific file
02Extract text, structure and layout
03Index document and section context
04Retrieve through MCP tools
05Return sourced context to agent