I'm an AI/ML & Data Product Leader with 21+ years of experience spanning technology, AI, data, and digital transformation across automotive and highly regulated financial services, including banking, lending, trust, and fintech. I specialize in product and enterprise AI strategy, turning ambiguous business problems into production-grade AI products, and owning the full lifecycle from vision, roadmap, and product discovery through deployment, adoption, and governance.
My expertise spans Generative AI, semantic search, retrieval systems, recommendation engines, Graph RAG, knowledge graphs, and agentic AI, combining deep product leadership with hands-on technical fluency. I am passionate about translating emerging AI capabilities into scalable products that deliver measurable business outcomes. Based in Tempe, Arizona.
I turn ambiguous ideas into AI products people adopt.
I'm a 0→1 builder. This portfolio shows the whole arc, spotting the opportunity, framing the problem, validating it, choosing the architecture, and shipping, through live demos and product case studies in financial services and automotive.
Where the judgment comes from.
Two decades shipping in regulated, high-stakes environments: banking, lending, and automotive safety. Now focused on applied AI.
Nine stages where ideas usually stall, and how I get through them.
Click any stage to see how I work through it, with real examples from financial services and automotive.
The first three stages determine everything that follows. Understanding customer pain points, framing the right problem, and defining the business outcome and KPI that matters most are where enterprise strategy and product strategy are won or lost. Models, architectures, and AI capabilities only matter after these fundamentals are in place.
Prototypes, not screenshots.
Each follows the same arc: the problem and the KPI that mattered, the ambiguity, how it was validated, the strategy and tradeoffs, the MVP, the architecture rationale, and the results.
AI Product Strategy, Architecture & Enterprise Adoption
Thoughts from building AI, ML, GenAI, RAG, Graph RAG, and Agentic AI products in regulated enterprises.
More articles
Upcoming articles on AI product strategy, enterprise architecture, governance, and adoption.
The frameworks I reach for.
Architecture is a product decision before it's a technical one. Three of the calls I make most often.
Should this be Rules, ML, GenAI, or an Agent?
| Approach | Choose it when | Example use cases (financial services) |
|---|---|---|
| Rules | Logic is stable, fully specifiable, and auditable. No learning required. | Eligibility cutoffs and hard declines, fraud velocity limits, regulatory disclosures, input validation and routing. |
| ML model | Enough labeled history exists and the task is narrow, repeatable, and measurable. | Credit risk and propensity scoring, churn prediction, fraud detection, document and complaint classification. |
| GenAI + RAG | Open-ended language over a body of knowledge that changes and must stay grounded. | Policy and product Q&A copilots, call-center assist, contract and call summarization, search over internal documents. |
| Agent | The task needs multi-step planning, tool use, and action across systems, with a human checkpoint. | Reconciliation and dispute handling, loan-package assembly, retrieve-then-act research, multi-system workflows. |
RAG, Graph RAG, or an LLM Wiki?
| Approach | Best when | Strengths | Watch-outs |
|---|---|---|---|
| Vector RAG | Answers live inside documents; questions are mostly lookup or summarize. | Fast to stand up, great for unstructured text, cheap to iterate. | Weak on multi-hop and joins; re-derives the same facts every query. |
| Graph RAG | Questions span relationships: "how is X connected to Y", exact counts, rollups. | Multi-hop reasoning, precise joins, and an explainable retrieval path. | Needs entity resolution and a modeled schema; more upfront design. |
| LLM Wiki | Knowledge is reused and evolving: the same entities answered again and again. | Consolidate once, update in place; stable, governable source of truth. | Investment in ingestion and freshness; overkill for one-off corpora. |