Solutions Architect — enterprise AI and agentic systems
15+ years across insurance, fintech, and SaaS platforms. I don't hand off the hard part — the same week I'm running a design review I'm usually also the one building the proof of concept and shipping the first production slice.
Current focus: multi-agent orchestration, RAG pipelines, and human-in-the-loop systems on Azure — built for observability, security, and governance from day one, with a strong bias toward keyless, credential-free architectures that remove secret sprawl at the infrastructure level.
| Project | What it is | |
|---|---|---|
| e-commerce-agents · docs | Six specialized agents over A2A — dual-stack Python and .NET on Microsoft Agent Framework, with generative UI | |
| ms-graph-mcp · docs | Microsoft Graph MCP server — 85 tools across M365, delegated OAuth | |
| mcp-generator · docs | OpenAPI to a governed, portable MCP tool surface — agent readiness for existing APIs | |
| ai-resources · docs | AI knowledge hub — fundamentals through architecture patterns, agents and LLM systems |
Production-grade starting points, each with a full documentation site.
| Project | What it is | |
|---|---|---|
| mean-docker · docs | MEAN stack reference — MongoDB 8, Express 5, Angular 22 (zoneless), Node 24, fully containerized | |
| clean-architecture-docker-dotnet-angular · docs | .NET 10 Clean Architecture with an Angular 22 signals frontend — RBAC, activity logging, PostgreSQL |
- Multi-agent systems on Microsoft Agent Framework and LangGraph — document processing, decision support, and process automation in enterprise contexts
- Agentic UIs with AG-UI and CopilotKit — conversational interfaces that expose agent state and reasoning to the user instead of hiding it
- RAG pipelines on Azure AI Search and Azure OpenAI, with retrieval evaluation, grounding verification, and content safety built in from the start
- Reference architectures for .NET Aspire + Angular/Next.js, following 12-factor and microservices principles
AI & Agents
Azure OpenAI · Microsoft AI Foundry · Azure AI Search · Microsoft Agent Framework · LangGraph · MCP · RAG · Document Intelligence · Anthropic Claude
Azure & Platform
AKS · Service Bus · Key Vault · Functions · App Service · Docker · Kubernetes · Helm · Managed Identity · Workload Identity Federation · Private Endpoints
Backend & Frontend
C# / .NET · .NET Aspire · Python · MediatR · FluentValidation · Serilog · Angular · Next.js · TypeScript · Tailwind
Data
Azure SQL · Cosmos DB · PostgreSQL · MongoDB
Security, Observability & Delivery
Zero Trust · keyless service-to-service auth · Entra ID · OWASP · multi-tenant isolation · OpenTelemetry · Application Insights · Azure DevOps · GitHub Actions
I use Claude Code for agentic development — scaffolding, refactoring, architecture exploration — and GitHub Copilot for in-editor completion. Every architectural decision and every line of production code goes through human review. The tools accelerate iteration; the judgment still comes from experience.
Most of what's here started as a problem I hit on a real project and then rebuilt in the open. If you're working on something similar — agent orchestration, keyless Azure, or an architecture worth thinking through — I'm glad to talk.
Blog · Projects · LinkedIn · X · [email protected]
I write about architecture, AI integration patterns, and lessons from enterprise delivery at nitinksingh.com.