Projects
Five case studies — what I built, and what happened.
AI-powered academic advising platform, built end-to-end from UI/UX through backend
integration with the Workday Student API and Okta SSO.
Led direct stakeholder engagement with Stevens' Dean of Undergraduate Academics —
gathered institutional requirements, shaped the roadmap around a real customer's
operating environment. Deployed and iterated on Vercel. Culminated in a fielded
leadership demo with potential reach across Stevens' 8,000-student population.
Sunset
AI-powered study platform for university students, onboarding 20–40 active student
users.
Market analysis surfaced a saturated landscape — well-capitalized incumbents
already serving the same need. Made the disciplined call to sunset at a clean
stopping point rather than force a crowded market.
BladePay
Sunset
React, Firebase, and Twilio SMS automation and payment-tracking platform built for
trade professionals.
Customer discovery with 30+ landscaping companies surfaced tight operating margins
as the core adoption barrier — the tool was wanted, the cost wasn't justifiable.
That finding drove the decision to sunset.
PL Converter
Built and deployed an internal AI-powered automation tool that eliminated manual
parts list processing — cutting engineering overhead and accelerating execution
across active Navy programs.
Defense
Vertical launch system integration supporting U.S. Navy surface combatants.
Full electronics design lifecycle — circuit and wiring architecture through
system-level integration for shipboard defense platforms. Direct collaboration
with military stakeholders, translating mission requirements into engineering
action.
Currently building
What's Next
In progress
An AI-assisted tool for mapping fields between a source document and a target
template, whatever format each one is in.
Grew out of PL Converter's internal automation work. Known formats match
instantly against saved profiles, and new ones get an AI-assisted first pass
that I review and confirm through a visual matching interface. Every confirmed
mapping is saved for next time, and each customer runs a self-hosted instance,
so no data or mappings cross accounts.