Icebreaker is a B2B social graph platform helping exceptional people discover and connect with each other. As co-founder, I lead design while contributing to product strategy, sales, marketing, and operations.
EVOLUTION
B2B Recruiting (Current)
Helping companies find and engage exceptional talent through their company's extended network, across LinkedIn, GitHub, X, Farcaster, and Icebreaker.
Job Seeker Focus
Ingested 1K+ top crypto companies and their jobs, multiplied by job seekers' existing networks to show who you know at which companies, facilitating warm intro hiring. Explored a recommend-a-friend flow based on DARPA balloon challenge, but found money taints referrals.
Onchain Professional Network
Built a robust multi-network search across Farcaster, LinkedIn, and Icebreaker itself, with a LinkedIn-style connection system and onchain credentials.
NFC Proof of Meeting
The key insight: meeting IRL is very hard to fake. Using NFC tech, we could prove people actually met in person. The challenge was getting people to change the social exchange dance.
REFLECTIONS
Startups are a game of perseverance. I've learned an absolutely tremendous amount beyond design — leadership, operations, marketing, GTM, fundraising. Painful amounts of learning. I've also learned a lot about myself, who I am, and what I care about.
Product-market fit is discovered, not designed. The hybrid product-design role keeps blurring — the future of building is being able to have the idea and execute it. With AI, execution is going to zero. It's a wild time to build stuff.
HOW I WORK
↳Trade-offs (what I cut + why)
- Killed the DARPA-inspired referral bounty system after finding that money taints the authenticity of referrals, the very thing our platform was built on
- Abandoned onchain credentials as a core feature when B2B customers made it clear they needed LinkedIn and GitHub integration, not wallet-based identity
↳Complexity → Clarity (making the tangled simple)
- Unified search across Farcaster, LinkedIn, GitHub, X, and Icebreaker into a single interface. Five different data models behind one search bar
- Distilled a multi-network social graph into a simple "who do you know at this company" view for recruiters who had zero patience for graph theory
↳Ambiguity (what I did when the brief was "figure it out")
- Phase 1 had no product spec. We started with NFC cards and a thesis that IRL meetings are hard to fake, then designed the product around what we learned at events
- Navigated four pivots without a roadmap, using each phase's failures as the brief for the next one
↳Systems thinking (upstream/downstream impacts)
- Designed the Chrome extension knowing that one recruiter syncing their LinkedIn graph would unlock network value for their entire company. Optimized for organizational adoption, not individual delight
- Built the "extended network" model recognizing that first-degree connections are table stakes. The real value is 2nd and 3rd degree paths to candidates
↳How I've changed
- Learned that product-market fit is discovered, not designed. I stopped trying to perfect features and started shipping experiments that would generate signal
- Went from a designer who needed a brief to a co-founder who writes the brief, sells it, funds it, and builds it
WEBSITE
icebreaker.xyz →METRICS
BACKED BY
MY ROLE
WORK

Lists
Collaborative list-building for B2B recruiting teams
View case study →
Chrome Extension
Fast LinkedIn network syncing for recruiters
View case study →
Onboarding
Designed and prototyped with Claude Code
View case study →
Brand
Visual identity evolution from V1 to V2
View case study →NFC Proof of Meeting
Using NFC to prove real-world meetings
View case study →