Work

Seven projects, each with the problem, what I did, and what changed.

Start with the index, filter by the kind of work you care about, or jump straight to a case. Where something is live, there's a link so you can check it yourself.

01 · Replit

Replit help center: knowledge that rebuilds itself every night.

  • Knowledge
  • AI adoption
Back to index ↑

Challenge

Replit ships constantly. Help content that lags behind the product doesn't just confuse readers; it feeds wrong answers to the AI assistant and to support.

What I led

As the first person building Replit's Learning & Knowledge function, I designed the approach: content as code, with one maintained source of truth instead of pages edited in place.

What I built

  • AI content generation using custom-built skills: pedagogy, writing, customer centricity, accessibility, and more…
  • A GitHub source that feeds the help center, AI support bots, courses and videos.
  • Drift detection that flags content falling out of step with the product.
  • People review every change before it publishes. The automated part is the nightly rebuild: approved source is rebuilt and republished every night.

What changed

Content that makes up about 2% of the docs became the #1 sourced content across searches. Ask AI used it 4–5x more than long-established pages.

#1
sourced content across searches, from ~2% of the docs
4–5x
more Ask AI usage than long-established pages

Go look at it: replit.com/help →

02 · Replit

An LMS in three weeks.

  • Learning & leadership
  • AI adoption
Back to index ↑

Challenge

Support onboarding needed a platform urgently. The conventional route, procurement then configuration, takes months before anyone learns anything.

What I led

I scoped what support onboarding actually needed and decided to build rather than buy, shaped weekly by the people learning on it.

What I built

  • Launchpad, a learning platform built from scratch, with a working beta in three weeks.
  • An AI practice engine to apply training skills with real scenarios.

What changed

The platform now runs support onboarding end to end.

Launchpad screen showing an AI-scored practice exercise
Launchpad: AI practice using real scenarios.
03 · Microsoft

Microsoft Office Training Center.

  • Knowledge
  • Learning & leadership
Back to index ↑

Challenge

The same everyday Office questions kept arriving in support, every one of them a ticket someone had to work.

What I led

I led the team behind customer-facing Office learning, choosing to focus on the 100 most-requested tasks and to measure behavior, not views.

What I built

  • Short, scenario-based videos on the 100 most-requested tasks.
  • A measure-and-improve loop based on what people actually did next.

What changed

350 million monthly users, and $120 million a year in support savings as people found answers instead of filing tickets.

Microsoft Office training page with short task-based video lessons
Office training content, still part of Microsoft support.
350M
monthly users
$120M
a year in support savings
04 · Microsoft

Microsoft Virtual Academy: 220K to 4.5M+ subscribers.

  • Learning & leadership
Back to index ↑

Challenge

Microsoft wanted free, global training for IT pros and developers, at a scale no classroom program could reach.

What I led

I led the team and the growth model: rapid, data-led iteration on what subscribers used, and steady expansion of the catalog.

What I delivered

  • A growth model driven by usage data, not catalog size for its own sake.
  • A continually expanding free course catalog for a global audience.

What changed

Subscribers grew from 220K to more than 4.5M in three years. The work earned the Bill Gates Training Visionary Award.

220K → 4.5M+
subscribers in three years
Award
Bill Gates Training Visionary Award

Want more detail on the growth model? Discuss the Virtual Academy project.

05 · Fortive

One standard for 400+ operating companies.

  • Change
  • Learning & leadership
Back to index ↑

Challenge

Fortive Business System training was fragmented across operating companies, and every acquisition added another version.

What I led

I set a single training standard and a scorecard to hold it, across 400+ operating companies.

What I built

  • Converted two to three days of classroom basics into three to four hours of self-paced learning.
  • Built AI translation for nine languages.

What changed

A year-long backlog cleared, engagements ran 40% faster, and translation costs fell 75%.

40%
faster engagements
75%
lower translation costs, across 9 languages

Project materials are not published here. Discuss the Fortive project.

06 · DoubleVerify

DoubleVerify: a four-tier leadership pipeline.

  • Learning & leadership
  • Change
Back to index ↑

Challenge

There was no consistent path from individual contributor to executive.

What I led

Working with external consultants, I designed a four-tier pipeline: aspiring leaders, new managers, managers of managers, executives, held together by a competency rubric. I also led the people workstreams for the acquisition of an AI company.

What I built

  • DV University.
  • Leadership 101.
  • A competency rubric for all four tiers.

What changed

Launched Leadership 101 and DV University; the privacy course reached 100% completion.

4 tiers
from aspiring leader to executive
100%
completion on the privacy course

Project materials are not published here. Discuss the DoubleVerify project.

07 · Independent

Six Second Resume: AI that drafts, people who decide.

  • AI adoption
  • Builder
Back to index ↑

Challenge

AI resume tools invent claims, and the person whose name is on the resume pays for it.

What I led

I designed the product around one rule: the AI drafts, the person decides what's true.

What I built

  • Built and launched a pipeline of specialist steps.
  • An independent truth check that flags unsupported claims for human review.

What changed

It's live at sixsecondresume.com.

Six Second Resume homepage
sixsecondresume.com, live.

Small bet first.

Hiring for a role, or need help with a project? Either way, let's talk for 30 minutes. If I'm not the right fit, I'll tell you who is.