Trust in AI & data

Confidence is easy. Calibration is the work.

Notes on AI and data systems, and the reasoning underneath them. Making something work once is the easy part. Showing that it holds, and knowing where it breaks, is what earns trust.

I write about the methods behind that trust: probability and decision science, machine learning concepts and formal verification.

Read the writing

What I work on

Thoughts

Logic, reasoning and the working-out behind a decision. These are experiences from real builds and how-tos worth sharing.

Evaluation & quality of AI

Metrics whose assumptions match the problem and assuring quality of the model. Doing the trajectory-level evaluation for agents. While focussing on the underlying data quality

Context engineering

Getting the right information in front of the model at the right moment, retrieval, grounding and knowledge graphs that turns scattered enterprise data into a queryable source of truth.

Integration

Wiring AI into the systems around it including use of tools, APIs, MCP and data pipelines. How does the demo scale?

Open-source & open-weight AI

Building on open models and open-source tooling. Control over cost, privacy and how a system actually behaves transparently.

Looking ahead

Where this is heading, which trends are worth taking seriously, the ones that aren’t and how the models and the tooling around them are evolving.

Read

Recent writing All posts