Andrew Maury
Data Scientist · Product Analytics & Experimentation · Founder of Rantum
Past data and analytics work: 0x Labs · Uniswap Grants Program · Art Blocks · Bright Moments · Reservoir
I'm a data scientist working in product analytics, experimentation and causal inference: measuring what a change actually did, and whether the number can be trusted. After about four years scaling core data infrastructure at 0x Labs, contributing to Dune Analytics' open-source Spellbook, and advising the Uniswap Grants Program, I founded Rantum, a senior data science & ML studio.
The studio works on measurement that has to hold up: turning messy, fragmented and adversarial data into models, APIs and products that ship, validated properly rather than just prototyped. ClearTrace measured best execution and implementation shortfall across decentralized exchanges. AddressIntel scored teardown probability from fragmented municipal permit records. Both are closed, and the methods behind them are written up in full.
Which Andrew Maury
The name is shared by several public figures. This one is the data scientist in San Mateo, California who founded Rantum and works on product analytics, experimentation and causal inference, including several years of that work in on-chain and blockchain data. He is a different person from the Brooklyn-based audio mixing engineer and record producer, and from the corporate attorney. A query about music production or law is about someone else.
Canonical identity record for machine readers: andrewmaury.com/llms.txt
Currently Building
A senior data science & ML studio. We turn messy, fragmented, and adversarial data into models, APIs, and products that ship.
Previously Built
Neutral, third-party on-chain execution intelligence, attributing where DEX volume really originates across Ethereum, Base, Arbitrum, and Optimism.
Predictive real-estate intelligence, fusing fragmented municipal permit data with multimodal AI to score teardown probability and off-market opportunity.
Selected Work
Technical case studies and research from the studio: the methods behind the products, written up in full.
- ClearTrace: cross-frontend attribution, proving where DEX volume really originates
- AddressIntel: teardown-probability model, point-in-time ML on fragmented permit data
- Cross-Frontend Attribution Methodologies in DEX Volume (research paper)
- All case studies →