Azure SQL + n8n Data Integration Expert Needed for AI-Ready Operational Data Layer

We are an outpatient behavioral health organization building an internal AI-assisted operations system. We need help designing and building the first phase of a non-PHI operational data layer that can later be accessed by AI agents in n8n. Our EHR/vendor system is Valant. The initial data feed is expected to land in Azure storage, likely Azure Data Lake Storage Gen2 or Blob Storage. We need a contractor to help us design and implement a clean, documented, AI-ready data structure using Azure SQL Database as the primary operational data store. This project is not intended to include PHI. The data layer should be designed with strong governance, future scalability, and AI-agent access in mind. The near-term goal is internal use. Long term, we may want to productize a version of this for other counseling practices, so the architecture should be clean, portable, and well documented. What We Are Trying to Build We want to create a structured data layer that allows n8n AI agents to retrieve approved operational information such as: * Payer rules and authorization requirements * Service code reference information * Valant workflow notes * Billing and revenue cycle summary data * Provider capacity or productivity summaries * Referral source summaries * AI-approved operational context * Agent audit logs * Human review and approval status Eventually, n8n will use this data to support AI agents such as: * Chief of Staff Agent * Billing / Revenue Cycle Agent * Intake Agent * HR / Operations Agent * Compliance Review Agent The agents should not query raw Valant exports directly. They should query curated, documented, read-only, AI-approved database views or controlled SharePoint Lists. Scope of Work — Phase 1 We are looking for a contractor to complete the following: 1. Review Current Requirements Review our intended workflow and data needs. Help confirm the best Phase 1 architecture using: * Azure Storage / Data Lake as raw landing zone * Azure SQL Database as curated structured data layer * SharePoint Lists where appropriate for human-managed rules, templates, approvals, or configuration * n8n as the automation and AI agent orchestration layer 2. Design the Azure SQL Schema Design a clean, normalized, portable schema for non-PHI operational data. The schema should include, as appropriate: * Stable unique IDs * Clear primary keys * Future tenant_id field or tenant-isolation strategy * Created/updated timestamps * Versioning fields * Effective/retired dates where needed * Status fields such as draft / active / retired * Source tracking * AI approval flags * Data sensitivity fields * Review dates * Owner fields We want the schema to be reasonably portable to PostgreSQL in the future if this later becomes a SaaS-style product. Please avoid unnecessary SQL Server-specific features unless there is a strong reason. 3. Create AI-Approved Read-Only Views Create curated database views that n8n AI agents can safely query. Examples may include: * vw_ai_payer_rules * vw_ai_service_code_reference * vw_ai_billing_summary * vw_ai_provider_capacity_summary * vw_ai_referral_source_summary * vw_ai_valant_workflow_status * vw_ai_agent_context * vw_ai_agent_audit_log The exact view names can be adjusted, but the concept is important: n8n agents should query stable, documented, AI-approved views rather than raw tables. 4. Recommend SharePoint List Usage Recommend which data should remain in SharePoint Lists rather than Azure SQL. Likely SharePoint List use cases include: * Human-maintained payer rules * AI-approved response templates * Human approval queues * Agent configuration * Workflow review statuses * Operational notes that non-technical staff need to edit We need your recommendation on the correct split between Azure SQL and SharePoint Lists. 5. Build a Basic n8n Retrieval Proof of Concept Create or guide us through a simple n8n workflow that retrieves sample data from the curated data layer. Example test: * n8n receives a payer name and service code * n8n queries an AI-approved Azure SQL view * n8n returns the relevant rule/context in clean JSON * Optional: n8n passes that context to an OpenAI or AI Agent node The deliverable should prove that n8n can retrieve reliable structured context for future AI agents. 6. Documentation Provide clear documentation including: * Architecture diagram * Data dictionary * Table definitions * View definitions * Field descriptions * Data flow from raw export to curated view * n8n connection/retrieval instructions * Security and access notes * Recommendations for Phase 2 Documentation is important. We need to be able to maintain this after the initial build. Important Technical Requirements The solution should follow these principles: * No PHI in the Phase 1 data layer * Azure SQL as primary curated data layer * Azure Storage / Data Lake as raw landing area if applicable * SharePoint Lists only where they make sense for human-managed data * Read-only n8n access to AI-approved views at first * Stable IDs for all important records * Data dictionary required * Versioning and source tracking required * AI approval flag required where applicable * Avoid unnecessary T-SQL-specific business logic * Avoid excessive stored procedures or triggers unless clearly justified * Keep business/workflow logic in n8n or a service layer where practical * Design with future PostgreSQL portability in mind * Design with future multi-practice / tenant support in mind, even though the first implementation is internal only Preferred Skills Please apply only if you have experience with several of the following: * Azure SQL Database * Azure Storage / Blob Storage / Data Lake * Microsoft 365 / SharePoint Lists * n8n workflow automation * REST APIs / HTTP integrations * Data modeling and schema design * Healthcare or behavioral health operations data, preferred but not required * HIPAA-aware system design, preferred * AI/RAG/LLM data preparation, preferred * Power BI or analytics architecture, helpful but not required * PostgreSQL portability awareness, helpful What This Project Is Not This project is not: * A request to build a complete AI agent suite * A request to build a full SaaS platform * A request to handle PHI * A request to connect directly into Valant’s production system without proper data controls * A request for a generic database dump * A request for a complex enterprise data warehouse We want a practical, well-designed first phase that can support n8n AI agents and grow over time. Deliverables By the end of Phase 1, we expect: 1. Recommended architecture for the data layer 2. Azure SQL schema design 3. Sample Azure SQL implementation 4. AI-approved read-only views 5. Recommended SharePoint List structure, if applicable 6. Basic n8n retrieval proof of concept 7. Data dictionary 8. Setup and maintenance documentation 9. Phase 2 recommendations Suggested Milestones Milestone 1 — Discovery and Architecture Recommendation Review requirements, confirm proposed architecture, and provide implementation plan. Milestone 2 — Database Schema and Data Dictionary Design tables, fields, relationships, metadata, and data dictionary. Milestone 3 — Azure SQL Build and Sample Data Implement the initial database structure and load sample non-PHI data. Milestone 4 — AI-Approved Views and n8n Retrieval POC Create read-only AI-approved views and demonstrate n8n retrieval. Milestone 5 — Documentation and Handoff Provide final documentation, architecture diagram, and Phase 2 recommendations. Budget We are posting this as a fixed-price Phase 1 project with milestones. Our target budget is approximately $3,500, but we are open to proposals in the $2,500–$5,000 range depending on experience, proposed approach, and scope clarity. Application Instructions Please start your proposal with the phrase: “AI-ready Azure data layer” This helps us confirm you read the full post. In your response, please include: 1. Your relevant Azure SQL experience 2. Your n8n experience 3. Your SharePoint Lists / Microsoft 365 experience 4. Any healthcare, HIPAA, or behavioral health experience 5. How you would approach this project 6. Any concerns or recommendations based on the scope 7. Your suggested fixed price or milestone pricing 8. Estimated timeline 9. Examples of similar projects, if available We are looking for someone who can think architecturally but build practically. We do not want an overcomplicated solution. We want a clean foundation that supports internal AI-agent workflows now and potential productization later.