Skip to main content

Work / RootData

RootData

I co-founded a crypto asset database and designed and collected its data from zero; it grew to 428K users and 200+ API partners, and I later opened the data to AI agents through MCP.

users
428K
API partners
200+
organic growth in a year
~14×
Role
Co-founder: database design, product, later advisor
When
2022–2026
  • Product management
  • MCP (TypeScript, Python)
  • Cloudflare Workers
  • XGBoost
  • Streamlit
  • SEO
Live: rootdata.com
22K+projects33Klinked wallets6,000+projects scoredRootData428K users700K–1M page views a daycharting, SEO, anti-crawling,CRM, wallet sign-inBinance AlphapartnerOKX BoostpartnerEthereum FoundationpartnerHashKeypartnerCoinGlasspartner200+ productsAPIMCP servers, ~20 toolsAI agents query it directly

What it is

RootData is a crypto asset database: projects, funding rounds, investors, people and the wallets behind them. I co-founded it in September 2022 and stayed until September 2026. The database was mine from zero to one: I decided what to collect, designed how projects, rounds, investors, people and wallets connect, and gathered the first data myself. Then I built the product on top and, later, advised.

It grew to 428K users, about 14 times over in one year without paid acquisition, with 700K to 1M page views a day, more than 22K projects and 33K linked wallets. More than 200 partner products, among them Binance Alpha, OKX Boost, the Ethereum Foundation, HashKey and CoinGlass, embed its data through the API.

What I built

  1. 2 monthsto 2,000 DAU

    A launch that held

    At launch I ran the product and the channels around it: 1,600 registered users, 2,000 daily actives and retention above 90% within two months.

  2. 0 → 1the database

    Designing the data before the product

    I chose what to collect and how it connects: projects to funding rounds, rounds to investors, people to the projects they built and the wallets they use. I gathered the first records myself, then turned the collection into a repeatable pipeline.

  3. 6product areas

    The parts users touch and the parts they don't

    Data preparation and charting, SEO, anti-crawling, UI and UX, the back-office CRM and wallet integration.

  4. ~20tools for agents

    Opening the data to AI agents

    Open-source MCP servers in TypeScript and Python, over stdio, SSE and Cloudflare Workers, so an agent can search projects, funding and investors as tools instead of scraping pages.

  5. 61models, AUC 0.81

    Predicting listings and launches

    A Streamlit analytics app with 61 XGBoost models over 6,000+ projects, chosen from 116 automated experiments, that scores which projects are likely to list on an exchange or launch a token.

See it running

The public site as it is today. The MCP servers are on GitHub.

Rootdata-MCP (TypeScript) and Rootdata-MCP-Python.

The RootData home page: market overview cards for market cap, ETFs, tokenized assets and prediction markets
Open the live site

Stack

Product
  • roadmap and specs
  • SEO
  • growth channels
  • partner integrations
AI
  • Model Context Protocol
  • TypeScript
  • Python
  • Cloudflare Workers
  • SSE
Data
  • XGBoost
  • Streamlit
  • the RootData API