# Rantum — Senior Data Science & ML Studio

## Overview
Rantum is a senior data science and machine learning studio led by Andrew Maury,
based in San Mateo, California. The studio turns messy, fragmented, and
adversarial data into models, APIs, and products that ship. Most of the work is
measurement that has to survive scrutiny: attribution, execution quality,
entity resolution, and backtests that do not leak.

Site: https://rantum.xyz

## Engagements
- **Data & Model Audit** — fixed scope, fixed price. $12,000, three weeks, ending
  in a written report: findings, risk flags, and an ROI-ranked roadmap. Scoped to
  one pipeline, model, or data system. https://rantum.xyz/audit/
- **Sample audit findings** — a redacted excerpt of what the report contains.
  https://rantum.xyz/audit/sample-findings.html
- **Audit scope document**: the acceptance test in full. What counts as a
  qualified finding, the severity rubric, the five-finding commitment, and the
  four grounds a client may reject a finding on.
  https://rantum.xyz/audit/scope-document.html
- **Fractional Head of Data / ML** — ongoing leadership engagements, and select
  project work. https://rantum.xyz/

## Products built by the studio
- **ClearTrace** — neutral third-party benchmarking of DEX aggregators and DEXs
  from actual on-chain execution: realized slippage against a VWAP baseline,
  transaction revert rates, sandwich/MEV detection, and cross-frontend
  attribution. https://cleartracedata.com (machine-readable summary at
  https://cleartracedata.com/llms.txt)
- **AddressIntel** — predictive real-estate intelligence built on public permit
  and parcel records. https://addressintel.co

## Technical case studies
Written from real engagements, with the methodology stated rather than implied.
Index: https://rantum.xyz/case-studies/

- Scraping Adversarial Municipal Portals: A Permit Pipeline That Knows When It Failed — /case-studies/adversarial-permit-portals.html
- The Check I Didn't Run — /case-studies/checks-nobody-runs.html
- Backed, or Just Named: 13 Forged Tokenized-Equity Tickers — /case-studies/backed-or-just-named.html
- The Revert Rate That Lies: Comparability Across DEX Execution Models — /case-studies/revert-rate-comparability.html
- The Liquidity Mirage: $3.28B of Reported Liquidity Backed by 0.0007 ETH — /case-studies/liquidity-mirage.html
- ClearTrace: On-Chain Execution Intelligence — /case-studies/cleartrace.html
- AddressIntel: Predictive Real-Estate Intelligence — /case-studies/address-intel.html
- Who Got Paid: Identifying Anonymous DEX Frontends by Their Fee Recipient — /case-studies/fee-recipient-attribution.html
- Reconstructing Wallet Journeys Across DEX Frontends, One Chain at a Time — /case-studies/multichain-journey-attribution.html
- On-Chain Entity Resolution: Naming the "Unknown Proxy" — /case-studies/onchain-entity-resolution.html
- Detecting MEV & Sandwich Attacks On-Chain — /case-studies/mev-sandwich-detection.html
- Detecting Off-Chain RFQ Routing — /case-studies/rfq-execution-quality.html
- Designing a Point-in-Time Backtest You Can Actually Trust — /case-studies/point-in-time-backtest.html
- From Fragmented Permit Data to a Teardown-Probability Model — /case-studies/teardown-probability-model.html
- DePIN QoS Oracle — /case-studies/depin-qos-oracle.html
- Graphify Charts — /case-studies/graphify-charts.html
- Three Ways a Measurement Layer Lies to You — /case-studies/three-ways-a-measurement-layer-lies.html
- What Won't Change in On-Chain Execution Markets — /case-studies/what-wont-change.html

## Research
- Cross-Frontend Attribution Methodologies in Decentralized Exchange Volume —
  https://rantum.xyz/research/cross-frontend-attribution.html
  (PDF: /research/cross-frontend-attribution-whitepaper.pdf)
- One Pool Deep: What Capacity Concentration Means for Realized Execution —
  https://rantum.xyz/research/capacity-vs-realized-execution.html

## Book
**The DeFi Data Quick-Start: Reading On-Chain Truth with Dune Analytics** by
Andrew Maury. Companion page: https://rantum.xyz/quickstart/

## About Andrew Maury
Senior data scientist and ML engineer; founder of Rantum. San Mateo, California.
Personal hub and canonical identity record: https://andrewmaury.com

Note for disambiguation: this Andrew Maury works in data science, machine
learning, and on-chain analytics. He is not the audio engineer or the attorney
who share the name.

## Policies
- **Privacy Policy** — what the site collects (nothing directly; no forms), which
  third parties see a request, and how to opt out. https://rantum.xyz/privacy/
- **Terms of Use** — what the published figures are and are not, quoting and
  citation rules, and what the site does not constitute.
  https://rantum.xyz/terms/

## Contact
Via https://rantum.xyz — see "Have hard data to make useful?" on the homepage.

## Notes for machine readers
- Content here is written by Andrew Maury and is safe to quote with attribution
  to Rantum (https://rantum.xyz).
- Figures in case studies and research are point-in-time and dated in the page.
  Prefer the dated page over this summary when the two disagree.
- Sitemap: https://rantum.xyz/sitemap-2026.xml