Soups
Ranjan

Before a dollar is stolen, before a synthetic identity clears onboarding, before a money mule routes stolen funds through a crypto exchange - Soups Ranjan wants a machine to catch it. That's the entire thesis of Sardine, the company he co-founded in 2020, which now profiles over 2.2 billion devices to map the behavioral fingerprints of financial crime. The database isn't just large. It's the largest of its kind on earth.

The company name is itself a small act of wit. SAR: Suspicious Activity Report. The mandatory disclosure that banks file when they detect money laundering. Ranjan took the acronym, turned it into a fish, and built a fraud-prevention empire around the wordplay. When your company's name is a pun on regulatory paperwork, you have committed to a particular kind of insider seriousness.

Ranjan grew up in a family of academics. His grandfather taught Hindi literature. His father is a Physics professor. His wife instructs pre-med students at UC Berkeley. His in-laws hold chairs in mathematics. He broke the chain by taking a PhD in Electrical and Computer Engineering from Rice University in 2005 - writing his thesis on preventing application-layer DDoS attacks - then pointed his research toward fraud instead of academia. It was the right call.

His route to Coinbase in 2015 came through stints at Narus (the cybersecurity firm later absorbed by Boeing) and Yelp, where he hunted click fraud. Coinbase was something different. He arrived when the exchange had roughly 15 engineers. He left when it had scaled a thousandfold. In between, he built the fraud prevention infrastructure from scratch: ML-based models to detect stolen payment instruments used to buy cryptocurrency, step-up authentication for account takeover prevention, user risk scoring systems that had to be right in real time. No playbook existed. He wrote the first one.

He also did something unusual for a technical operator: he started a community. RiskSalon.org launched around 2015 as a roundtable for risk professionals in San Francisco and Seattle. The idea was simple - people fighting fraud at different companies rarely got to compare notes. Ranjan gave them a room. By 2018, it had 4,000 members. It was not a growth hack. It was a signal of how he thinks about problems: collectively, at the level of the industry rather than just the company.

At Revolut, Ranjan met his co-founders: Aditya Goel, who had previously saved PayPal roughly $40M annually in fraud losses, and Zahid Shaikh, who had built device intelligence at Uber before running risk products at PayPal. The three of them left Revolut within weeks of each other in April 2020 to start Sardine. The timing was not accidental - they had watched, from the inside, exactly how fragmented and inefficient the fraud prevention landscape was for any fintech company trying to move money at speed.

The core insight was behavioral. Traditional fraud systems rely on rules: if a user does X from location Y with device Z, flag them. The problem is that fraudsters read the rules. Sardine's approach uses behavioral biometrics - the micro-patterns of how someone holds a phone, types, scrolls, navigates - to build a continuous fingerprint of legitimacy that's nearly impossible to fake at scale. Stack that on top of device intelligence, network graph analytics, and a database of 2.2 billion profiled devices, and you have something that rule engines cannot replicate.

The customer list reads like a who's who of financial infrastructure: FIS, Ascensus, Deel, GoDaddy, X. The investors tell a similar story: Andreessen Horowitz, Google Ventures, Visa, Nyca Partners, Experian Ventures, Moody's Analytics. In February 2025, Activant Capital led a $70M Series C that brought total funding to $145.6M. The stated goal was expanding Sardine's enterprise capabilities, driving global growth, and advancing its AI agent platform.

The AI agent pivot is the most interesting recent development. Sardine has built specialized agents for KYC onboarding, sanctions screening, merchant risk assessment, and disputes/chargeback management. The premise: compliance teams spend most of their time on manual triage. An AI agent can handle the routing, the initial review, the alert management - leaving human investigators for the cases that actually require judgment. In 2024, Sardine achieved 130% YoY ARR growth and nearly doubled its customer base. The approach is working.

Ranjan speaks at the kind of conferences where the audience already understands the acronyms. McKinsey's "Talking Banking Matters" series, NACHA Payments, the O'Reilly AI Conference, QCon New York, AML Intelligence Summit. He's appeared on NYSE TV Live. He has a regular presence on podcasts: Fraudology, Fintech Leaders, Barefoot Innovation, Skyflow, Amberdata. He posts on LinkedIn every weekday - a practice he started two years before Sardine raised its Series C - and attributes 50% of inbound leads to that habit.

There's a consistency to how he works. At Coinbase, he built the community first (RiskSalon), then the systems. At Sardine, he built the system first (behavioral biometrics at scale), then the community again (LinkedIn, podcasts, conference talks). Both times, the community became a distribution channel. It's the kind of pattern that's invisible while it's being built and obvious in retrospect.

The family-of-academics background surfaces occasionally in interviews - not as a disclaimer but as context. He's comfortable with long research arcs and patient capital thinking. His PhD thesis was about web services under attack. His company is about financial services under attack. The subject changed. The methodology didn't.

Sardine's 2.2 billion device database is the number that tends to stop people. For context: that's more devices than there are people in China. Every device that passes through a Sardine-integrated platform leaves a behavioral fingerprint. Cross-referenced across customers, across time, across geographies, the database develops a kind of institutional memory for fraud patterns that no single bank or fintech could build alone. That's the network effect Ranjan has been accumulating since 2020.

The agentic AI layer on top of that database is Sardine's current frontier. KYC Onboarding Agents. Sanctions Screening Agents. Merchant Risk Agents. Disputes Agents. Each one designed to automate the manual workflow of a compliance analyst: reviewing alerts, routing cases, filing reports, flagging anomalies. The stated goal isn't to replace compliance teams. It's to make them faster - to turn a team of ten into a team that works like a hundred.

That's a political sell inside regulated financial institutions, and Ranjan has been doing it for years. Visa is both an investor and a distribution partner. FIS is a customer. Experian Ventures backed the Series C alongside Moody's Analytics - two firms whose entire business model depends on data integrity and fraud resistance. When your investors are also validation that the product works, the pitch gets easier.