Start with the user outcome
I begin with what the product should help someone accomplish, then let that outcome shape the interface, architecture, and validation.
AI shapes both the product
and the way I build it
I run an independent AI product lab where agentic workflows span product direction, design, architecture, implementation, testing, and deployment. The output is a portfolio of AI-native products and the systems required to build and operate them.
One lab across products and their foundations
Some projects are AI-native products for creative work and conversation intelligence. Others are user-facing products in security and search, or capabilities inside larger products, including identity, authorization, connectors, and shared platform work.
VectorCI and Vector Code sit on the build and operations side of the lab. I created them when existing CI and agent development workflows introduced more cost, friction, or loss of control than I was willing to accept.
AI is also part of how I work. I use agents across the product development loop while retaining responsibility for the product decisions, technical boundaries, and evidence behind the result.
How the lab operates
Use AI across the whole loop
I use agentic workflows across product direction, design, implementation, review, testing, deployment, and operations.
Build the missing foundations
When existing tools, integrations, or infrastructure limit the product, I treat that gap as part of the work.
Prove it in use
I keep human authority visible and validate what the complete system does inside the environment where it has to work.
One lab from user products to supporting systems
Every project is either the product, a capability inside it, or a system I use to build, validate, and operate it. Together they show how I move from product intent to working software.
View selected work