Work that shipped (and what didn't)

Here's what I've built, why we made certain bets, what moved the numbers, and what I learned when things didn't go as planned.

Three themes run through everything

  • AI that actually ships (not just impresses in demos)
  • Enterprise translation (B2C to B2B needs more than features)
  • Systems under pressure (forecasting, compliance, time trackingβ€”not the most exciting work, but where money flows)

Career timeline

From aerospace engineering to AI products that ship. Each chapter links to the work below.

Filter:
πŸ›‘οΈ

Protexxa Defender

The Enterprise Pivot

Senior AI Product Manager (Leading Product Function) β€’ 2024–2025

B2B SaaSEnterprise PivotAgentic AISOC 2ComplianceMLOpsPLGCross-functional Leadership

OUTCOME

  • β€’ 25,000 users in 6 months (enterprise + district deployments)
  • β€’ 40% reduction in analyst workload
  • β€’ SOC 2 certified (unlocked enterprise deals)
πŸ€–

FastDoc

From Zero to Marketplace in 6 Months

Founder & Product Lead β€’ 2024

Next.jsAtlassian ForgeLLMsGenAIFounder0β†’1

OUTCOME

  • β€’ 100+ companies in 6 months
  • β€’ 70% reduction in documentation time
  • β€’ 30% activation increase
✈️

Solution Tek

The Product That Never Launched

Product Manager β€’ 2017–2020

B2B SaaSForecastingRevenue OptimizationAviationPredictive Analytics

LESSON

  • β€’ $12M+ in annual revenue potential validated
  • β€’ Pilot-ready platform built alongside the flagship product
  • β€’ COVID paused the entire industry
πŸ’Ό

Tempo Timesheets

Scaling a Mature Product

Product Manager (Timesheets & Planner) β€’ 2021–2024

MLSaaSProduct ManagementAnalyticsScale

OUTCOME

  • β€’ +12% ARR growth
  • β€’ NPS 38β†’52 (customers finally felt heard)
  • β€’ 20+ long-requested features shipped

The work that didn't ship

Learning comes from shipping and failing. Here are the misses that shaped the bets that followed.

The AI Feature Nobody Used

  • Context: At Tempo, we built an AI-powered "smart scheduling" feature that could optimize team capacity across projects.
  • Why we built it: the data was there, the algorithm worked, user research said people wanted it.
  • What happened: 3% adoption. Teams ignored it.
  • Why it failed:
  • - Solved a planning problem when execution was where the pain lived.
  • - Required too much setup (calendars, priorities, constraints).
  • - Felt like it was managing people, not helping them.
  • What I learned: AI products fail on adoption, not accuracy. Better question: what's the smallest behavior we can automate that people already do manually?

Documenting the next one

  • There's always another experiment in the backlog. I'm documenting the next one now.
  • If you're curious about the messier drafts, let's talk.

Want help shipping the next chapter?

If you're navigating an enterprise pivot, scaling a mature product, or validating an AI bet, I can help you connect the dots between strategy and execution.

Email meContact page