A governed knowledge layer that exposes KPI, schema, product, and semantic context through a reusable MCP tool surface.
Impact
one governed tool surface for product, KPI, schema, definition, and relationship context
Hiring signal
MCP server design, enterprise retrieval, multi-index search, identity-aware Azure integration, and reusable knowledge-platform architecture.
My role
Lead developer · solution architecture, MCP tool surface, orchestration backend, product-aware index resolution, multi-index retrieval, and Azure identity integration
Status
Development-stage architecture and trace.
The problem
Enterprise knowledge is distributed across KPI dictionaries, schema references, semantic relationships, definitions, documents, and product-specific metadata. Without a shared access layer, each AI workflow reimplements retrieval logic and receives inconsistent business context.
My approach
The solution is organized into content-management, knowledge-store, and MCP-access layers. Curated definitions, KPI dictionaries, table and column metadata, semantic relationships, and product-specific context are indexed in Azure AI Search. A remote MCP server exposes this knowledge through focused Azure AI Foundry agents for product, KPI, schema, and relationship resolution.
Architecture judgment
MCP provides a versioned, reusable contract for multiple consumers. Specialized agents keep product, KPI, and schema resolution independently testable, while the product registry selects the correct indexes without requiring downstream client changes.
Architecture
The complete execution path, service boundaries, data stores, security controls, validation stages, and operational telemetry.
Implementation
Technology
Development trace
Run the trace to see the main workflow stages.