From Google Fraud Fighter to Cancer Trial Matchmaker

What we talked about.

This episode of the Sunny Ray Show features Augie, a Croatian born, Waterloo trained computer engineer who spent 13 years at Google building payments infrastructure and later leading AdWords risk and integrity engineering. After his father's prolonged cancer battle, Augie pivoted into life sciences, joining Grail to work on early cancer detection through blood based screening. He later co-founded Betteromics, a multi-omics data platform for life sciences later acquired by Apple Health, before starting Lind AI three years ago. Lind, named after James Lind, who ran what is considered the first controlled clinical trial in 1747, is a research platform that helps hospitals qualify patients for clinical studies by automating the burdensome data discovery work traditionally done by research nurses. Augie discusses how skills from fighting fraud at planetary scale transferred to finding rare cancer signals in noisy biological data, why roughly 80 percent of clinical trials fail to meet enrollment timelines, and how Lind aims to move patient access to trials from about 4 percent today toward 100 percent.

A former Google fraud fighter turned cancer researcher explains how AI could get every patient offered a clinical trial.

The questions, and the answers.

What is Lind AI, at a high level?

Lind AI is the company I founded three years ago. It is a research platform for hospitals. We help qualify patients for clinical studies and help research operations identify more opportunities and treatment options for patients on the ground, taking on 90 to 95 percent of the data discovery workload that research nurses and staff normally have to do manually.

Why did you decide to build this specifically?

Over the last 12 years in life sciences I saw firsthand that the biggest impediment to clinical research is qualifying and finding enough patients to test therapeutics. Research nurses and staff face burdensome data discovery workloads, so we built a platform to handle most of that work on their behalf.

What did 13 years at Google, including payments and AdWords risk work, teach you about trust and human behavior?

Building Google Checkout taught me that you can be an innovator inside a big company, not just a startup, and we learned the hard way what it takes to win people's trust while dealing with fraud and abuse. Running AdWords risk and integrity for seven years showed me that any marketplace is really about brokering trust, and even low volumes of fraud can create outsized damage if you are not vigilant.

What pulled you away from Google toward something with higher stakes?

My father went through a prolonged cancer battle with numerous procedures and clinical trials, and that forced me to pause and think hard about what I wanted to do next. I felt compelled to pivot into life sciences and bioinformatics, so I joined Grail to lead data science and AI development for early cancer detection using cell free DNA.

How did your background in payments, risk engineering, and machine learning prepare you for healthcare?

A lot of the AI and data science techniques carried over well. Fraud and abuse detection is actually similar to cancer detection because both involve finding rare, low frequency events in noisy data. The modeling challenges, like overfitting on noise, were similar even though the domains, biology versus payments, were completely different.

Why did you name the company Lind, and what does it say about your mission?

We are named after James Lind, who is credited with the first modern clinical trial when he tested citrus as a treatment for scurvy using multiple interventional arms and a control group. We wanted to pay homage to that foundational work while acting as an accelerant for clinical research today, since human biology still makes running studies onerous, tedious, and error prone.

Roughly 80 percent of clinical trials fail to meet enrollment timelines. When did you realize patient matching, not the science, was the real bottleneck?

I learned this at Grail while running studies of different sizes. Many trials shut down due to poor patient accrual or underperforming sites, and often the population enrolled is biased toward large academic centers and more advantaged patients, which means approved therapeutics can underperform in the general population later. That gap is what Lind is built to close.

Can you explain what happens for a cancer patient without Lind versus with it?

Normally, whether a patient gets considered for a clinical trial depends on whether their physician happens to know of one, often from memory, a journal, or a conference. With Lind, every patient can be pre-screened automatically against all candidate trials before their appointment, so the clinician walks in already equipped with treatment options. Today only about 4 percent of patients are ever offered a study, and we want to get that number close to 100.