Product Executive · Builder · Technologist
I build enterprise products that move at scale — and AI tools that actually solve problems. Twenty years of shipping things that matter, from $100B contract platforms to Toastmasters dashboards.
About
I started as a software engineer at Fidelity Investments, crossed into financial services at Morgan Stanley, and eventually found product management — the place where technical depth, business strategy, and human insight converge. It turned out to be the sweet spot I'd been circling my whole career.
What's always driven me is something simpler than the résumé suggests: taking processes that are slow, manual, and fragmented, and turning them into streamlined, intelligent systems that move at the speed businesses need. For twelve years at Agero I did that for connected vehicle and roadside assistance platforms serving 115 million consumers. At Amazon and AWS, the problems got bigger — GenAI products, $100B+ contract lifecycle systems, agentic intake workflows, and investment platforms managing billions in annual programmatic funding. Five years later, I'm looking for the next challenge at the VP or Director level.
What I do outside of work tells the other half of the story. I build AI-powered tools — dashboards, automation agents, analytics platforms, clinical prototypes — because I can't help it. The best way to understand where AI is headed is to build with it. Currently exploring ideas at the intersection of geolocation and AI for a project I'll share when it's ready.
AI Projects
An autonomous agent that fetches live DCP data for all 138 Toastmasters clubs across District 31, classifies each club's health using time-aware business logic (on-track, watch, at-risk, critical), drafts personalized outreach emails to Area Directors, and produces a District Director briefing report — all in a single run. Includes a three-pattern eval framework: deterministic test cases, rule-based structural checks, and an LLM-as-judge layer that scores every email draft for specificity, tone, and actionability before it goes anywhere near a real officer.
Why I built this
"District leadership was manually tracking 138 clubs across spreadsheets and Slack messages, then writing individual outreach by hand. I designed a system that does in one run what used to take days — and built the eval framework because I wanted to know, not just hope, that the outputs were trustworthy."
An agentic prototype for wealth management advisory practices that transforms raw meeting transcripts into fully orchestrated post-meeting workflows. A two-tier trust architecture governs every action: Tier 1 (internal, non-client-facing) executes autonomously with a full audit trail; Tier 2 (client-facing or regulated) batches into a single advisor review point. Regulated items — custodian forms, contribution instructions — always require individual confirmation regardless of trust phase. Every action is sourced back to the exact transcript moment that generated it.
Why I built this
"I wanted to think rigorously about where AI should and shouldn't act autonomously in a regulated environment. The architecture isn't just 'approve or reject' — it's about what trust earns: less friction inside a safe zone. It never earns a bigger safe zone."
A pre-visit clinical briefing agent for primary care, built around a care graph and RAG retrieval layer working together — the graph handles relational data (which labs came from which referral), RAG handles unstructured visit notes. Includes a companion evaluation layer that scores every generated briefing for faithfulness, relevance, and completeness before it ever reaches a clinician, plus dedicated tools for HCC coding review and patient risk stratification. Four deployed AI prototypes, all backed by real Anthropic API calls against a synthetic patient population.
Why I built this
"I wanted to understand, hands-on, what it actually takes to make AI trustworthy enough for a clinical setting — not in theory, but by building the evaluation layer myself and watching it catch a real hallucination during testing. That one catch taught me more about AI trust than any article on the subject."
More Projects
A fully agentic job search workflow built on Claude Cowork. Drop a JD into a folder — it scores across five weighted dimensions, generates a tailored ATS-optimized résumé if the score clears 75, and archives low-fit roles automatically. Daily schedule, zero manual intervention.
18 verified economic and social metrics for every president from FDR through Trump II. Primary sources only — BEA, BLS, Census, OMB. Outcomes, not verdicts.
Statistical comparison of Federer, Nadal, and Djokovic across 12+ dimensions — Elo ratings, surface simulations, serve/return metrics, head-to-head. ATP Tour data, primary sources throughout.
Five-step Facebook post workflow fully automated — Google Sheets → QR code → Pillow flyer compositing → Facebook Graph API. Browser-based, zero technical knowledge required for end users.
Eight-tab dashboard tracking all 138 District 31 clubs — auto-CSV ingestion, year-over-year comparison, snapshot trend charts, drill-down modals. Used by district leadership weekly.
Quarterly incentive tracker for 138 clubs — Up Your Game streak logic, Membership Building, and Pathways Level Up rankings from Education Achievements data. Club drill-down modals throughout.
Fully automated role and attendance tracker for Network Voice Toastmasters. Meeting agenda PDFs in Google Drive trigger a GitHub Actions workflow that parses roles, matches members, and updates a live dashboard — zero manual data entry.
Experience
L6 Technical Product Manager — AWS Investments & Alexa
Built mission-critical financial systems at AWS scale — a contract lifecycle management platform automating Private Pricing Agreements, GenAI intake and routing for the Generative AI Innovation Center, funding and deal-shaping tools for AWS Growth Initiatives, and a two-way sales program matching console used by the entire AWS field sales organization. Also spent time in the Alexa org building the ML data preparation platform used by 3,000+ scientists.
Principal Engagement Manager II
Managed delivery for two tier-1 P&C insurance clients on a smartphone-based driving data platform, covering distracted driving detection, usage-based insurance, crash and claims, and mobile sensor data processing.
Product & Program Management Consultant
Took over a troubled $20M program to migrate Zipcar's reservation platform across North America and the UK. Directed 40 developers, 7 product managers, and 3 UX designers. Delivered the final London market migration on time and under budget. Also led the company-wide transition to Agile, moving 150 product and engineering employees to Scrum within 6 weeks.
Senior Director, Product & Program Management
Twelve years building enterprise products for one of the largest connected vehicle and roadside assistance platforms in the US — serving 115 million consumers in partnership with auto manufacturers and insurance carriers. Delivered a direct access insurance integration that saved $2.7M annually and lifted client satisfaction from 91% to 95%. Identified and resolved a critical billing flaw recovering $10M–$12M annually in previously lost revenue. Spearheaded 10+ connected vehicle integrations with global automotive OEMs and pioneered a first-of-its-kind Visual IVR platform.
Software Engineer · Fidelity Investments (2001–2005) · Financial Advisor · Morgan Stanley (2005–2006) · Financial Consultant · New England Financial (2006–2007)
Started as a software engineer at Fidelity, helping build a $30M B2B platform for registered investment advisors that saved $32M in annual costs. Then crossed into financial services — first at Morgan Stanley, then New England Financial — managing a $3M book of brokerage and retirement assets across investments, retirement planning, estate planning, and insurance. This combination of engineering depth and financial fluency has shaped every product decision since.
Skills
Speaking & Training
Spoke on a GenAI panel at an Amazon-wide conference for engineers, TPMs, PMs, and leaders across the company — sharing perspectives on practical AI product development, agentic workflows, and where enterprise AI is heading.
A one-hour approachable introduction to AI for senior residents at an assisted living facility — no jargon, no fear. Covered letter writing, understanding medical terms, recipe help, and live Claude demos. Designed from the ground up for a non-technical audience.
On making AI accessible
"If you can type a question, you can use AI."
A 60-minute virtual workshop for NYC metro Toastmasters club officers — turning AI tools into real content leverage. Covered magnetic marketing content, NotebookLM podcast generation, video creation, bio polishing, and LinkedIn presence. Every attendee left with a working prompt toolkit.
The session's opening line
"AI doesn't replace your voice. It amplifies it."
Two training sessions for club officers across Massachusetts and Rhode Island — practical, hands-on guidance on using AI tools like NotebookLM, Claude, and others to reduce workload, create better content, and run more effective clubs.