Senior .NET engineer. I build AI agent systems with engineering discipline — verification gates, evals and measurable quality instead of demo magic.
15+ years across manufacturing, legal and banking domains; six of them leading teams. I care about the boring parts that make software last: domain modelling, testability, and the harness that lets an application keep evolving safely.
- AI systems for the SDLC — support tooling, integrations with the systems teams already use, and what it takes to run all of it at scale.
- The agent harness — skills, automation, evals and verification gates around coding agents. I test the tooling hands-on: Claude Code, GitHub Copilot, whatever ships next.
- Distributed .NET — CQRS/DDD, microservices, .NET 10 / Aspire.
I write a field-guide blog on building reliable AI agent systems — agent skills, multi-agent orchestration, feedback loops, and designing an agentic SDLC with honest measurements. Every claim is tagged by source strength.
| Project | What it is |
|---|---|
| Warehouse | Microservices WMS in .NET 10 / Aspire — DDD, event-driven, a full test pyramid, ADRs and CI/CD. |
| RuleCraft | You define an interface; an LLM implements it from a plain-language spec at runtime — verified, human-approved, hot-loaded. No redeploy. |
| mkasperczyk.com | Writing on agents, architecture and the harness around them. |
- Backend — C# / .NET, Aspire, CQRS/DDD, microservices
- AI — LLM integration, LangGraph / agent orchestration, RAG, prompt & eval design
- Data — Databricks, Spark, Python
- Front end — React, TypeScript (jQuery → Angular → React, monorepos)
- Cloud & delivery — AWS · Azure, Azure DevOps, CI/CD
From planning through production to demos. Process follows the work, not the other way around — Scrum, Kanban, back to Scrum once teams run smoothly.
AWS Certified: Solutions Architect – Associate · Developer – Associate