Start with the consequence
Ask what needs to become better for a user, team, or business before choosing the model or feature.
About
I learned product judgment in systems that were already live.
Before working on AI products, research, and patents, I spent years inside software, release, networked, and broadband systems where a wrong assumption could become a production problem. That experience shaped how I think: a useful idea has to survive the system, the operating reality, and the people who have to use it.
The connection
Reliability, product, research, and IP are not separate tracks.
Reliability work teaches you to distinguish an observation from an explanation. It also teaches you that a technically correct answer can still be useless if nobody knows what to do next.
I apply that habit to AI product work. Start with the workflow and consequence. Decide where a model helps. Make the weak points visible. Give people a sensible way to review, override, or stop the system when needed.
Inventorship adds another lens. A patent forces precision about the system being claimed. A prototype tests whether the idea can become a working artifact. Research tests whether the underlying technical proposition survives a more formal evaluation.
A product belief
Not every AI problem needs more AI.
Some problems need a better model. Others need a clearer workflow, stronger controls, better observability, or human judgment at the right point. Knowing the difference is product work.
How I approach the work
Three operating habits
Design the failure path
Observability, escalation, rollback thinking, and explicit limits are product behavior, not cleanup work.
Keep the claim inspectable
Patent status, manuscript status, prototype limits, and operating scope should be easy to check and difficult to misread.
Professional arc
The path to applied AI product work
2014–2018
Software, QA, and systems engineering
Enterprise integration work on the Massachusetts Health Exchange platform followed by QA, software-development, release-validation, and production-readiness work in connected-device environments.
2018–Present
Release operations, platform reliability, and field triage
Broadband platform work spanning release readiness, feature rollout, reliability analysis, and, in the current scope, telemetry-driven field triage and customer-impact investigation.
2024–Present
Independent AI product research and inventorship
Patent-backed systems, prototypes, and research across deterministic remediation, provenance, replay, reliability, and bounded AI execution.
2025–Present
Doctor of Engineering in AI · Penn State
Formal study alongside active research and product development, building on graduate education at the University of Pennsylvania.
Work
See operating work
Field triage, release support, reliability, and the operating patterns behind the work.
Patents
Inspect product systems
Issued and filed work presented with current status and public-safe context.
Research
Review technical questions
Accepted and under-review work across replay, provenance, runtime architecture, and agentic AI.
Insights
Read the product thinking
Direct links to selected writing and technical briefings behind the portfolio thesis.
What I am looking for