AI Systems Architecture • Product Engineering • UX
I build products end to end — from the problem, through the system, to the interface people actually touch.
I own product development end to end: framing the problem, designing the system, and writing the software that ships. My current focus is adaptive AI systems — products that decide what to do from live signals rather than handing everyone the same fixed answer.
Selected work
An adaptive AI training system
A training system that keeps the plan aligned with what is actually happening — goals, training activity, recovery, measurements, consistency and progress evidence — so adaptation is the product's job rather than the user's.
Read the case study
How I work
I get specific about who is struggling, with what, and what would count as evidence that the product helped. Scope follows from that answer instead of arriving before it.
I work through how the pieces relate: what owns which decision, what moves between them, and where the seams need to be visible when something behaves unexpectedly.
I write the product, not only the plan — interfaces, application logic, data model, and the connective work between them. Building it myself keeps the architecture honest about what is achievable on a real schedule.
Adaptive behavior is only valuable if it is inspectable. I define the boundaries an AI-driven system runs inside — what it owns, what it may decide, what it must declare when it changes course — and build the validation that proves the behavior is the one I specified.
Edge cases, empty states, and failure modes are part of the product. I ask what happens when data is missing, the network is slow, or someone does something reasonable that I did not plan for — and I build the checks that catch it.
Architecture matters because of what the user feels: fewer surprises, clearer feedback, and interfaces that stay legible under real conditions rather than only in the happy path.
Contact