Back to Projects
Cloud - Data Pipeline
Document Ingestion & Embedding Pipeline
Reliable document ingestion and embedding generation for enterprise RAG.
- Role
- AI Engineer
- Timeline
- Accellor
- Stack
- Azure Container Apps, Blob Storage, SharePoint
Problem
Enterprise corpora change constantly. Ingestion had to stay current without reprocessing everything, and failures needed safe retries.
Solution
Scheduled Container App Jobs pull from SharePoint, ADI, and Blob Storage, generate embeddings, and apply delta/retry logic for resilient updates.
Technology Stack
Azure Container AppsAzure Blob StorageSharePointPythonDockerGitHub Actions
Learnings
Delta detection and retry semantics matter as much as the embedding model when the corpus is large and continuously changing.