Topic
Agentic AI and developer tooling
End-to-end agents, reusable skills and evaluation loops — built, published, and used on real engineering corpora.
Building agents instead of talking about them: role-based playgrounds, deterministic entity linkers, standards query services, and the scaffolding that makes an AI coding assistant work to a stated standard.
Experience
Where it was practised
Projects
What got built
Agents Kindergarten
A testable playground for agent roles, reusable skills, evaluation, and improvement loops.
Creator and maintainer
ASPICE agentic QA framework
Multi-agent generation and gap analysis of automotive process evidence, with functional-safety cross-referencing.
Creator and maintainer
glassgraph
A deterministic entity linker for research wikis, built as the opposite trade-off to LLM extraction.
Creator and maintainer
Filogram
Making a corpus of roughly 3.3 million engineering files navigable by clustering rather than by search alone.
Creator and maintainer
RepoElements
A periodic table for software systems, keeping measurement separate from interpretation.
Creator and maintainer
Reasoning lenses
Lenses that reason the way a specific person or field reasons, forged from a real corpus and auditable move by move.
Creator and maintainer
Claude project template
A project scaffold that makes an AI coding assistant work to a stated standard from the first commit.
Creator and maintainer
SpyMeet
A meeting transcription pipeline built for noisy audio, technical jargon, and more than one language.
Creator and maintainer
Content analytics pipeline
Measuring what published technical writing actually does, instead of remembering what it felt like.