AI

AI Agent Hub

Multi-agent pipelines for structured & unstructured data. Compliance-first, cost-controlled.

Architecture
AI Agent Hub Architecture
Active

AI-powered ETL (structured data)

Auto-ingestion → dbt generator (models, tests, personal-data map from the source schema) → Control-M / legacy jobs converted to Airflow

Sources
CSV / ExcelParquetAirbyteDatabase
Agent Flow
Ingest Agent→Compliance (PII)→Quality Check→Transform (dbt)
OUTPUT: dbt models + tests + personal-data map + Airflow DAGs, in your own repo
Auto-ingestion (Airbyte) →Generated dbt project + column lineage →Orchestration (Airflow) →Control-M → Airflow converter →
Active

Document AI Pipeline

PDF, DOCX, images, HL7/FHIR → parse, extract, embed → RAG chat with sources

Sources
PDFDOCXImagesHL7 / FHIR
Agent Flow
Parse + Extract→PHI Scan→Embed (LanceDB)→RAG Chat
OUTPUT: Vector knowledge base + Q&A with source citations
Open Pipeline →