About me

I build production AI systems, and I care whether they behave the way they should and why they do what they do.

After 20+ years in tech, shipping software, architecting cloud infrastructure, and leading engineering teams, I’ve spent the last few years watching generative AI move from research curiosity to production reality. And daily I run into questions like:

• Why does swapping a pronoun change the response?
• Why was the word “dead” chosen for this README.md file?
• Why does a model trained to be helpful sometimes not be?

Questions like those have pulled me toward AI safety and responsible AI.

What I’m working on

My current focus is the gap between a GenAI system that demos well and one that’s actually trustworthy in production. That means:

• Building with AWS Bedrock, Guardrails, and agentic architectures, and watching closely what breaks and why
• Studying mechanistic interpretability through the ARENA curriculum and running experiments with the Claude API to understand model behavior from the inside
• Participating in AI red teaming exercises (NIST and Humane Intelligence) to find the edges of what systems will do
• Applying explainability tools like SHAP in production ML contexts

I also write and speak about this work. I’ve presented on AI explainability at dev meetups and at Tusculum University, and I share experiments and resources publicly on GitHub.

I love collaboration

My learning doesn’t exist in a vacuum. I’m a strong believer in collaboration and knowledge sharing. That’s why I’m excited to connect with other data and AI enthusiasts and learn from their experiences.

Let’s connect on LinkedIn to share insights, learn from each other, and push the boundaries of responsible AI.

My favorite personal project

“Hello Penguins” GitHub repository, a collection of machine learning experiments with the Palmer Penguins dataset. Inspired by the “Hello, World!” programming tradition, this repository is a series of small experiments to illustrate foundational machine learning concepts. Each experiment includes evaluation metrics and visuals to verify the model predictions make sense and are explainable. I love that it is a mix of software engineering, data science, and (inexpensive) cloud hosting.

My specialties

I enjoy using technology to solve problems, caring for data, and using AI coding assistants to learn as much as possible and increase my development velocity.

Now page

What I’m up to right now.