Skip to content

Applied research

We investigate engineering problems that emerge when systems must become reproducible, explainable, and operable in production.

Three study topics

Each topic keeps its own question, systems, and publications together. Results from one program are not mixed with another.

01In progress

Reliable generative pipelines

Engineering Reliable Generative Pipelines

Generative imaging is used as a demanding workload to investigate data provenance, reproducibility, resumability, inference contracts, and distributed compute.

How can an experimental generative pipeline become a reliable, testable system?

02Ongoing work

Knowledge as Code

How to represent organizational facts in versioned files, then validate their consistency automatically before they become stale.

How can organizational knowledge become machine-verifiable?

Public system

Cueto

Versioned organizational facts validated by a compiler in CI.

View the study
03Ongoing work

Inspectable software systems

How to build automation whose decisions, priorities, and rationale remain visible and controllable by engineering teams.

How can a decision be automated without creating a black box?

Public system

Pratrol

Pull-request triage with inspectable decisions and rationale.

View the study

One shared discipline

An explicit question, a documented protocol, inspectable artifacts, and honest limits. That is what turns an experiment into engineering evidence.

View all Blog publications

Investigating a similar question?

We can frame the decision, build the protocol, and connect the results to an operable architecture.

Discuss your architecture