Azure RAG Recommendation Platform Build

I’m looking for an experienced Azure-native team to design and implement an enterprise-grade product-recommendation platform based on LLM powered Retrieval-Augmented Generation. The solution must live entirely in Azure, drawing content only from our existing PostgreSQL product database (read-only) so that every suggestion is grounded in real, up-to-date catalogue data.

Core expectations
• Azure-first architecture: App Services, Azure AI Foundry components, and a secure data layer deployed through ARM/Bicep or Terraform.
• Intelligence layer: semantic matching with Azure AI Search combined with orchestrated calls to OpenAI (or an equivalent) through Azure Cognitive Services and Azure Machine Learning pipelines.
• Clean API surface: a documented REST/GraphQL layer that separates business logic from presentation, so web, mobile, and future channels can all consume the same recommendation engine.
• Conversational UX: optional chat endpoint or Bot Framework hook that reuses the same orchestration layer.
• Security by design: Azure AD authentication, granular RBAC, managed identities, Key Vault secret storage, and audit logging enabled from day one.

Key deliverables
• High-level and component-level architecture diagrams
• Infrastructure-as-code for all Azure resources
• RAG pipeline code, fine-tuning or prompt-engineering artefacts, and unit tests
• REST/GraphQL documentation (OpenAPI/Swagger) plus a sample chat interface
• End-to-end CI/CD workflow in Azure DevOps or GitHub Actions
• Operational playbook covering monitoring, logging, and rollback procedures

Acceptance criteria
1. Recommendations reference only catalogue entries that can be traced back to our PostgreSQL DB.
2. Average recommendation latency ≤ 1 second for the top-10 results.
3. All endpoints pass penetration testing with zero critical findings.
4. Deployment is repeatable from a clean Azure subscription using the supplied IaC scripts.

If you have proven experience with Azure Cognitive Services, Azure Machine Learning, Azure AI Search, and large-scale API builds, I’d love to review your approach, timeline, and examples of similar RAG solutions you’ve delivered. Show More