Sam Cusano
I build 0→1 products and the systems underneath them.
For 14+ years, I’ve worked across AI, marketplaces, and enterprise software, usually when the brief is fuzzy, the technology has opinions, and everyone has a theory.
My job is to make those theories collide with something real.
Something that tells us whether we’re onto something or just having a very expensive conversation.
Here are the receipts.
- Revenue impact
- $7M+
- Design systems built, scaled, or contributed to
- 7
- Companies
- 8
210
GitHub contributions in the last 12 months
An enterprise billing system. A marketplace revenue rebuild. An AI product from zero to one.
Three very different plots

Apartment List
Could six weeks rebuild trust across 20% of at-risk ARR?
- Impact: 20% of at-risk ARR recovered
- Timeline: 6 weeks

Leafsight
Could a proposal be done before the next job?
- Impact: $300k ARR
- Timeline: 6 weeks

Preqin
What if the data found the insight before an analyst did?
- Impact: 90% faster time-to-insight
- Timeline: Zero to one
Ship early. Let reality decide.
I’d rather have a bad prototype than a good argument.
When a decision gets stuck, I build the smallest thing that can make the disagreement concrete.
Sometimes it breaks immediately. Great. We learned something.
Sometimes it works, and now everyone has something real to react to.
A prototype gives the conversation something to push against.
So, what do you want to know?
Complex product design, mostly 0→1 products and ambitious rebuilds.
AI workflows, B2B SaaS, marketplaces, and the occasional problem nobody quite knows how to describe yet.
A senior or principal IC role building something hard.
I've led design end to end with PMs and engineers. I've picked up product strategy when PMs leave. I've worked across the fuzzy bit where design becomes a product decision.
Usually by asking the annoying question nobody wants to answer yet.
I kept asking it until the team started asking it without me. That's generally the goal.
Sequencing beats capability.
A model might be able to do ten things. If someone catches your AI being wrong twice, they're thinking, can I trust this thing?
Plenty.
A leaderboard I was convinced would work that made people feel exposed. An MVP I scoped beyond what the timeline could carry.

