My Journey in AI

Welcome! I’m Yael,

 

DESIGN → IMPLEMENT → SYSTEMATIZE → TEACH → ADVISE

I build AI systems where data, knowledge, and reasoning come together.

I’m Yael Bernstein, a Senior AI Engineer at Northern Arizona University.

​

My path into AI engineering has been unconventional. I spent nearly 20 years working deep inside enterprise systems, data, governance, institutional reporting, and the business processes those systems support. Much of that work sat where technical architecture, institutional knowledge, data quality, and accountability intersect.

​

That background shapes how I approach AI.

 

I am interested not only in what a model can do, but in what has to exist around it for the system to be useful and trustworthy:

• How do you give an AI system access to the right data?

• How does it understand what that data means?

• How do you represent the terminology, concepts, relationships, rules, and institutional knowledge that people inside an organization already carry?

• How do you build that system so it can be governed, evaluated, reused, and trusted?

 

Those questions have increasingly become the center of my work.

 

From data to knowledge

 

Beginning in January 2025, I entered an intensive period of technical development across Azure, Microsoft Fabric, data engineering, AI engineering, retrieval, agents, and the evolving Microsoft AI ecosystem.For 20 months, I pursued an unusually concentrated period of technical development across cloud, data engineering, Microsoft Fabric, AI engineering, agents and knowledge systems. 

​

I spent weekends and holidays in Microsoft Learn, technical courses, Fabric Dojo, Reactor sessions, documentation, certifications, and hands-on experimentation. At the same time, I was applying those concepts directly to university data and AI projects.

That period gave me the technical map I was looking for.

​

What I brought with me was just as important: years of experience translating complex institutional rules into data structures, validating high-stakes data, tracing how information moves across enterprise systems, and understanding where definitions, ownership, quality, and governance determine whether a system can be trusted.

​

Today, I am much less interested in collecting credentials and much more interested in building.

My work is increasingly focused on the layer between enterprise data and AI reasoning: the semantic structure that allows an AI system to understand not only what data exists, but what it represents and how concepts relate to one another.

​

That has led me deeper into metadata, taxonomy, ontology, knowledge graphs, retrieval, and governed knowledge systems. I am also building code-first ontology pipelines so those semantic models can be created, validated, versioned, and evolved as part of an engineering workflow rather than treated as a one-time modeling exercise.

​

What I am building:

Repeatable, 

For our Data Agent work at NAU, I have been building a governed technical knowledge system around the implementation process itself: current source documentation, structured Markdown knowledge, provenance, lifecycle metadata, implementation guidance, evaluation practices, and reusable engineering patterns.

The goal is straightforward:

Every validated piece of engineering work should make the next piece of work easier.

​

How I tend to work:

​

Building teh system around the system 

​

​

​

Where I am now: 

Data engineering · AI agents · semantic knowledge · ontology · retrieval · knowledge graphs · governance · Microsoft Fabric and Foundry

​

The Accelerated Learning Years

January 2025 → September 2026

​

For 20 months, I pursued an unusually concentrated period of technical development across cloud, data engineering, Microsoft Fabric, AI engineering, agents and knowledge systems. The goal was never the certifications themselves. It was to build the technical foundation required to design and implement enterprise AI systems.

By September 2026, that phase had done its job.

The next phase is about building.