Case Studies

Real problems, real solutions, real results.

Azure Functions

Bulk Dataverse Import Pipeline

Problem: 20,000-row Excel files requiring weeks of manual data entry into Microsoft Dataverse for a government client.

Solution: Azure Function on Azure Government handling Excel parsing, parent record lookups, and $batch inserts at 1,000 operations per request. System-Assigned Managed Identity for auth (no credentials in code), per-row error logging back to SharePoint for full audit trail.

Result: 40,000 individual API calls reduced to roughly 40. A process that would have taken weeks was fully automated and repeatable.

Azure Functions · C# · Dataverse Web API · Managed Identity · SharePoint
AI Integration

DelveMUD AI Integration Layer

Problem: Building AI features (NPC conversation, dungeon generation, quest creation, item crafting) that need to stay up under real load without creating vendor lock-in or a single point of failure.

Solution: Multi-provider LLM abstraction layer where each AI subsystem (NPC agent, dungeon director, quest generator, item appraiser) runs independently behind a provider interface. Per-subsystem rate limiting, caching, and fallback chains ensure resilience. No game-side code needs to change when a model is swapped.

Result: Models are hot-swappable with zero game-side changes. The architecture pattern translates directly to production enterprise AI features.

Go · Anthropic Claude API · Microservices · Rate Limiting · RAG Patterns