Delegate the mechanical tasks to agents. Keep the intelligence.
Tell me about your project.
01 · Foundation
Start with your systems.
AI is a powerful tool, and a risky one. It only helps when it works with the systems, the decisions and the people you already have.
- Field
- Software engineering, agentic AI and enterprise platforms (Dynamics 365, Power Platform): 5+ years of enterprise software delivery.
- Sectors
- Public sector, healthcare, automotive, real estate, luxury and fashion, industrial distribution.
- Clients
- Delivered
- Agentic systems on enterprise platforms, and PCF components, including one integrating ArcGIS for the public sector.
02 · Instruments
Give intelligence something to work with.
I build the instruments agents need to work on real enterprise systems: tools, clear context and shared memory. Instruments, not engines.
- Skills
- Agentic engineering and agentic workflows shaped to each project, MCP servers, context engineering and memory for agents and teams, human-in-the-loop governance.
- Delivered
- An agentic delivery framework with human checkpoints, used across teams since March 2026.
- Delivered
- A context and memory layer for agents.
- One example
- MCP Dataverse, an open-source TypeScript MCP server, listed in the official MCP Registry.
- 100 tools
- 28 categories
- 8,000+ npm downloads
03 · Delivery
Move forward. Keep people in control.
From the first experiment to a system in production: agentic delivery with a human checkpoint at the decisions that matter, proof at every step, and releases your team can review and run.
- Skills
- Agentic delivery with human checkpoints, release governance, ALM and CI/CD on Azure DevOps, architecture review.
- Delivered
- In healthcare, a redesign of ALM and CI/CD.
04 · Transmission
Build a team that can continue.
The handover is part of the work. Teams need to understand, question and adapt what they receive.
- Skills
- Enterprise AI adoption, training teams in AI, handover, shared context for people and agents.
- Delivered
- Teaching, mentoring and architecture review with teams of 5 to 30+ engineers.
- Mentoring
- In English, one to one or in small groups: the concepts first, then how to apply them.