Work
Things I've built
Work
Rosvelt
2026Most teams use tickets to track work. Rosvelt uses tickets to ship work.
You write what you want built. An AI agent scopes it, plans it, codes it, tests it, and opens a PR. You review and merge. That's it.
The idea came from building too many products for other people. I kept doing the same thing: take a ticket, understand the codebase, write the code, test it, ship it. The process was repetitiv
Comcast
2024-25Worked with Fleming Shi, CTO of Nitel (acquired by Comcast), on migrating a telecom platform for enterprise clients.
Holdex
2023-24Worked as a product developer on Ozean, Clearpool and Truflation. Products with more than 500K users, backed by Sequoia and other VCs.
Projects & Ideas
Lindora
Virtual try-on for clothes
Lindora
2026Online fashion has a dirty secret. Between 30% and 40% of everything purchased gets returned.
That's not a bug. That's the business model. Brands already price in the returns. We pay for them. And the root problem is simple: you're buying based on how clothes look on someone else's body. A model. A mannequin. A flat lay photo. Never you.
I watched my girlfriend return item after item. She'd open
Monako
Animated faces for AI agents
Monako
2026Every AI agent needs a face.
Not a complex 3D avatar. Not a static icon. Something in between. Something that blinks, looks around, feels alive. Something you can drop into any project without thinking about it.
I kept needing this for different projects. A face for a chatbot. An avatar for a dashboard. A character for a demo. Every time I'd either use a boring circle with initials, or waste hou
VOA
Voice AI infrastructure
VOA
2026I was integrating voice into Lindora when I realized: I'd done this before.
Set up ElevenLabs. Connect OpenAI. Wire up audio streaming. Build conversation state. Handle edge cases. The same 2-3 weeks of work. Every. Single. Time.
Every product that wants voice has to build the same things. Audio capture and streaming. Speech-to-text pipeline. LLM integration. Text-to-speech synthesis. Conversati
Paxie
Prediction markets without crypto UX
Paxie
2025Prediction markets fascinate me. Not for the money. For the information.
Markets aggregate what thousands of people know into a single number. Better than polls. Better than experts. Better than algorithms. If you want to know if something will happen, ask people to put money on it.
The problem: every prediction market I tried felt designed for crypto people, not normal people.
Here's what usin
Hoosling
Crypto crowdfunding, my first startup
Hoosling
2021My first real startup. I was 16.
The idea was simple: let anyone invest in startups using crypto. No accredited investor requirements. No geographic restrictions. Just a wallet and a dream.
Traditional startup investing is broken. If you want to invest in a promising company early, you need to be an accredited investor (meaning you already have money), know the right people, live in Silicon Vall