About
About me
I'm Luiz Claudio Hogrefe, a software architect based in Cologne.
I publish as Luiz Hogrefe. I have been building software since 2008, first in Brazil and, since 2014, in Germany. Most of those years were spent in enterprise projects where an architecture has to survive audits, legacy systems and handovers, not only a first release.
Today my work sits where software architecture meets governed AI, semantic interoperability and traceable digital evidence. I care less about what a system can do in a demonstration than about what it refuses to do, what it can prove afterwards and who is allowed to change it.
How I got here
I started in 2008 as a programmer at Edusoft in Blumenau, working on the financial and academic routines of a web-based academic management system. At FURB, the regional university, I worked as a software developer and later as a systems analyst, and I completed my Bachelor's degree in Computer Science there. My thesis applied ontologies to the development of information systems in a cloud-computing setting, a subject that still runs through my work on semantics.
In 2014 I moved to Germany. After an internship in Java development in Berlin and a role as a software developer at a private-cloud provider in Düsseldorf, I worked independently on web and mobile applications.
From 2017 to 2022 I was a consultant at Sopra Steria, in projects for energy, banking and automotive clients. The work ranged from developing web applications to the architecture and documentation of an orchestration platform and deployment concepts. After a year in Brazil in the interim management of a family business, I returned to consulting at msg in Cologne, as developer, Scrum Master and test manager.
In 2024 I began independent research on semantic interoperability, digital product passports and the European rules that will require them. In 2025 I worked at Accenture in quality engineering, as Scrum Master and tester in an enterprise programme. From August 2025 to April 2026 I architected and led the development of AnyImob, an AI assistant for real-estate brokerage. From May to August 2026 I was a software architect in the Standards division of IW Consult in Cologne, working on architecture and data models for digital product passports and on ECLASS as semantics for digital twins.
Since September 2026 I have been developing my research independently through AnyLAI, a non-commercial R&D project.
How I think about architecture
For me, architecture is the set of decisions that are expensive to reverse, written down where they can be checked. A diagram helps. A boundary that the build pipeline enforces helps more.
In regulated environments this becomes concrete. A requirement from the ESPR, the EUDR or the AI Act is not a label to attach to a system afterwards. It is an input, like latency or data volume, and it has to show up in data models, interfaces, review steps and in the evidence a system keeps.
Capable AI models change where the difficult work is. When implementation gets cheaper, intent, constraints and verification become more valuable. That is the thesis I am working on now.
How I work
Five principles I apply to my own work. They are engineering habits, not laws of nature.
- 01
Architecture over slogans
A system is described by its boundaries, its interfaces and what it refuses to do. A claim that cannot be traced to one of those is not yet architecture.
- 02
Deterministic boundaries around probabilistic systems
Models are useful and fallible. They do bounded work behind rules that decide, keep a record and can be replayed, and they are never where permission comes from.
- 03
Evidence someone else can verify
A result is worth what a reader can check without trusting its author. Its sources, its version and the person responsible for it travel with it.
- 04
Standards without lock-in
Public standards first, such as W3C Verifiable Credentials, UN/CEFACT and GS1, so that what I build can be checked, replaced or handed over without me.
- 05
Make ambiguity explicit before automating it
Automation amplifies whatever it is given. An ambiguous requirement becomes a consistent mistake, so I write the ambiguity down and resolve it with the people who own it first.
Languages
I work in four languages, which is also why this site exists in four.
- Portuguesenative
- GermanC1
- EnglishC1
- Spanishprofessional working proficiency
Current research
My research is developed and published through AnyLAI, an independent, non-commercial R&D project. It is not a company, and nothing it publishes is an offer.
So far it includes COADF, a published framework for building AI-assisted software under explicit governance; ERQYO, a runtime architecture for governed decisions with a reference implementation; and AnyDPP, a research environment for evidence-based digital product passports.