Project successes

The projects others found too hard.

A crowded market with no budget. A product buried under conflicting requests. Government data that wouldn't reconcile. A process no one could scale. Here's what happened — each told as the challenge, the approach, what I did, and the result.

Growth · Crowded MarketConsumerStartups

1,000% growth in a market called unwinnable

A Absolute Movers (Optimize Marketing)

The challenge

A one-year-old, cash-only moving startup in a market with 260+ entrenched competitors and almost no marketing budget — the kind of fight most would call unwinnable.

Approach

Don't outspend anyone — compete on trust. Pinpoint what customers actually feared (hidden fees, fake insurance, no-show movers) and arm them to tell the honest companies from the rest.

What I did

Built a customer-education campaign — a “15 questions to ask any mover” guide and an evidence packet (real insurance, bonding, guaranteed price) that tactfully exposed competitors' gaps and shifted the decision from price to integrity.

Result

1,000% sales growth in 90 days, an 85% phone conversion rate, direct-mail response up 300% — sold out within 12 months.

Product LaunchHealthcare & EHR

A new product line grown 3500% to $18M

Leprechaun LLC — Medicare Advantage / HCC

The challenge

A healthcare-IT firm needed a brand-new product and a genuine competitive edge to win Medicare Advantage plans — territory no one internally had cracked.

Approach

Find the wedge: let buyers see the financial upside before they commit.

What I did

Defined and launched “Prospective,” wrote the MRDs/PRDs that drove engineering, and built a predictive model used in the pre-sales cycle to forecast each plan's financial benefit.

Result

Grew the product 3500% — $500K to $18M in two years — and lifted deal close rates by 100%+.

AI Data · GovernmentGovernmentEnterprise SaaS

Reconciling 8 data sources to lift tax revenue 36%

RippleNami — Uganda government

The challenge

Eight disparate, inconsistent government data sources had to be reconciled to identify tax non-compliance — a politically and technically fraught $3M program.

Approach

Treat identity and trust in the data as the real hard problem, and architect for it.

What I did

Led the system end to end — identity resolution, fuzzy matching, and advanced RBAC on a modern cloud microservice stack — managing a 24-person team.

Result

Increased tax revenue 36% within the first 7 months of GA (the largest quarter ever) and delivered 30% under budget.

Transformation · CPOHealthcare & EHREnterprise SaaS

An operationally stuck platform, unstuck

Evexia Diagnostics — 30+ lab SaaS aggregator

The challenge

An e-commerce lab platform was operationally stuck: most new clients abandoned enrollment, there was no CRM, and growth was capped by manual work.

Approach

Remove the friction that was quietly killing conversion and partnerships.

What I did

As Chief Product Officer: redesigned the enrollment flow, stood up CRM and help-desk from scratch, rebuilt the partner program, and simplified the UI from 25+ menus to 5 tabs across 60+ Figma screens.

Result

Registrations rose from 33% to 92%, integration partners grew 325%, and revenue per employee climbed from $800K to $1.2M on 30%+ growth.

AI AgentsHealthcare & EHRStartups

66% more revenue from one person, via AI agents

AI workflow automation (medical-lab e-commerce)

The challenge

A manual spreadsheet ran the enrollment pipeline and matched lab results to orders — a bottleneck everyone tolerated because automating it looked too messy.

Approach

Let AI agents do the repetitive judgment work, with humans only where it matters.

What I did

Designed an AI-agent workflow that managed pipeline status, sent auto-email replies, and matched lab results to orders end to end.

Result

Supported a 66% revenue increase with a single team member and no added headcount; registrations climbed from 33% to 92%.

AI BuildStartups

Building Code Wizard — an AI product built hands-on

Optimize Consulting (in development)

The challenge

Building codes are fragmented, dense, and slow to navigate — a genuinely hard domain to make trustworthy with AI.

Approach

Build it hands-on to prove the business⇄tech bridge, not just specify it.

What I did

Designing and building an AI-powered building-code assistant with modern AI tooling (Claude Code, Vercel v0, Replit).

Result

In active development — a working demonstration of AI-agent product building end to end.

Have one that's stuck?

Bring me the goal and the constraints — especially the parts that make it hard — and I'll help you ship the result.

Start a conversation