I did my PhD at UCSB under Andy Maul in educational measurement. Early in grad school, I devoted time to studying quantitative methods in educational research - focusing on psychometrics and causal inference. I also dove into exciting-to-me debates about test validity theory and philosophy of science.

As I joined more projects in grad school, it often seemed like conceptual issues masqueraded as statistical or data-oriented questions. I had a creeping feeling that I could contribute more by helping gain some conceptual clarity around defining the things we intended to study and identifying when researchers (including me) were being bewitched by our words..

At the same time, there was burgeoning work comparing and contrasting traditions of measurement in the human sciences and the field of metrology (measurement sciences outside of the social sciences) that captured my interest - particularly taking advantage of concepts already developed in metrology like the concept of a measurand.

These interests turned into a job at NWEA where I worked on an R&D team in ontology and measurement developing cognitive models in service of classroom-based assessments. After NWEA was purchased by HMH, my work became more technical - working as an ontology engineer, deploying enterprise-level ontologies and serving machine learning systems.

What might an ontology engineer do?

Good question - here is a rough and broad example:

1. Coordinate with subject matter experts (like learning scientists) in developing conceptual clarity around the things relevant to an organization (or scientific field) and how these things relate to each other.
2. Work with consumers of the ontology (e.g. machine learning engineers, data scientists, data engineers, etc.) to consider how the ontology should be structured.
3. Move these semantics into a machine readable language (such as the Web Ontology Language).
4. Align data to this ontology, linking data across fields, ideally reducing ambiguity around internal vocabularies or what an organization’s data refer to.
5. Developing design patterns, data models, and querying methods enabling machine learning engineers, data scientists, or business analysts to retrieve data based on relevant relations between data.