Naftali
Harris

Sometime around 2014, a PhD candidate at Stanford packed up his statistics research and took a call from Max Levchin. Levchin, co-founder of PayPal and then building Affirm, needed someone who could think hard about risk - the kind of risk that doesn't show up cleanly in textbooks. Naftali Harris took the job. He became Affirm's first data scientist. And in doing so, he walked directly into the world's fastest-growing financial crime.

Synthetic identity fraud - the practice of stitching together real and fabricated personal information to create a "Frankenstein identity" - was spreading through U.S. financial systems like a slow infection. Harris saw it up close. Traditional fraud detection tools weren't built for it. The credit bureaus weren't equipped for it. The banks largely didn't know how to name what was happening to them. He did.

In 2017, Harris co-founded SentiLink with Maxwell Blumenfeld - another Affirm alumnus - and brought a statistician's rigor to a problem that had been treated mostly as an operational nuisance. SentiLink was built around a specific idea: that identity fraud is a data problem, and data problems have solutions if you're looking at the right signals.

The Company He Built

SentiLink's core product is real-time fraud detection for financial institutions opening new accounts. When someone applies for a credit card, a personal loan, or a digital wallet, SentiLink's system runs the application against its identity and risk models, surfacing a score that indicates whether the applicant is who they claim to be. What makes SentiLink different from legacy fraud tools is the specificity: it doesn't just flag anomalies, it detects synthetic identities - manufactured people who may have perfect credit histories built over years.

The numbers tell a spare story. Synthetic identity fraud accounts for roughly 85% of all identity fraud cases in the United States. Institutions lose approximately 5% of annual revenue to fraudulent activity. SentiLink currently verifies more than three million identities every day. Its customers include 10 of the 15 largest U.S. banks, five of the top 10 credit unions, and over 30 fintech unicorns. That's not a niche product; that's infrastructure.

Funding and the Path to Scale

SentiLink closed its $70 million Series B round in August 2021, led by David Sacks at Craft Ventures, with participation from Felicis Ventures, Andreessen Horowitz, and NYCA Partners. The round brought total funding to $84 million and brought Sacks and Victoria Treyger onto the board as observers. The company has since grown to nearly 200 employees across its San Francisco headquarters and beyond, with Harris personally based in Austin, Texas.

The funding trajectory mirrors a broader truth about identity verification: as more financial services move online, the attack surface for synthetic fraud grows. SentiLink has expanded from a single product into a broader suite - including the Identity Theft Score, which won the 2026 FICO Industry Vanguard Decision Award - while maintaining the real-time scoring architecture that made it valuable to banks in the first place.

Before SentiLink: A Statistician in the Wild

Harris arrived at the University of Chicago as a statistics undergraduate. After graduation, he continued at Stanford as a Graduate Fellow in the statistics PhD program - a program he eventually left, famously noting on LinkedIn: "When I was 22 I dropped out of grad school..." The ellipsis does a lot of work there. What came after was a data science internship at Kaggle, where he designed predictive modeling competitions, followed by Twitter, where he implemented a new machine learning algorithm that ran 15 times faster than its predecessor. Then Affirm. Then SentiLink.

His personal website still carries the same tagline it has for years: "Statistician, Hacker, and Climber." The rock climbing is not metaphorical. Harris is an actual climber, and the disposition it takes to find routes up blank faces - patience, pattern recognition, an unwillingness to quit when the path isn't obvious - maps cleanly onto how he built a company in a space that most established players had written off as intractable.

Code for Its Own Sake

Harris maintains an active GitHub profile with projects that have nothing to do with SentiLink's core business. There's Tauthon, a fork of Python 2.7 that backports Python 3 syntax and builtins for developers unwilling to migrate. There's a Python implementation of abstract algebra - group theory from scratch. There's Markovian, a chess engine built as a weekend project. There's a geographic calculation he ran to determine his home's latitude and longitude purely from measuring the angle of the sun at noon.

These are not portfolio pieces. They're the output of someone who finds computation genuinely interesting for its own sake - which is, it turns out, exactly the personality type you want running an identity intelligence company. The fraud patterns SentiLink hunts are not static. They evolve. They require people who treat problem-solving as an intrinsic activity, not a deliverable.

Recognition and What It Signals

The industry has taken notice. Forbes named Harris to its 30 Under 30 Finance list in 2025. American Banker recognized him as one of its Innovators of the Year in 2024. He served as a Forbes 30 Under 30 judge in 2026. SentiLink's Identity Theft Score - which quantifies the risk that a real person's identity has been stolen and used in an application - won the FICO Industry Vanguard Decision Award in 2026. The company achieved FedRAMP Ready status in May 2025, opening the door to federal government partnerships.

Harris speaks at RSA Conference, LendIt Fintech, and FinCrime forums. He has appeared on CNBC - once with Deirdre Bosa discussing the SentiLink founding story, and again in a Kate Rooney piece that introduced "Frankenstein identities" to a mainstream business audience. His presence in these venues isn't promotional; it reads more like someone trying to explain a technical problem to an audience that needs to understand it.