Forward¹

Forward¹ Link to heading

(Note: short form, interactive version originally posted on my LinkedIn; check it out at this link.)

After five and a half wild years at NVIDIA, today I left to go start the next phase of my career. I could not be more excited but I will also miss so many people - I got to work with some of the kindest, most brilliant people I’ve ever met. Much is (now) written about NVIDIA, and something that’s talked about but hard to capture is its “One Team” ethos. That attitude driving 30,000+ people is something truly special to be a part of. Succeeding and failing only together makes it all way better.

A lot happened and a lot has changed since my first interview in 2019. Back then, Voyager, the current crown jewel of the headquarters buildings, was just a sketch and a hole in the ground. The company itself was one third its current size, the Life Sciences team I joined was 5-8 people depending on how you counted, and I was a bright-eyed but slightly burnt out PhD student about to defend.

Coming out of my PhD at University of Cambridge and the National Institutes of Health, no one could understand why I would go to a gaming chip company. Several people I talked to pronounced the company “nuh-vidia” and warned me that by going to a gaming company I would be squandering my career prospects as a computational scientist, and while I would collect a decent paycheck, I would also fade into obscurity.

Well… while I knew they were wrong, I was off by a few orders of magnitude.

Even in 2019 NVIDIA was showing accelerated genomics code at their American Society of Human Genetics booth. A stroke of misfortune for Fernanda Foerrter (getting trapped in the airport due to thunderstorms) meant when I showed up to the Baylor Structural Variation Hackathon, the NVIDIA team needed a leader, and Ben Busby “volunteered” me to head it. I loved working with the team and was floored to learn NVIDIA was very, very actively contributing to open source bioinformatics. We gelled instantly, and I still consider the people on that hackathon team friends.

Coming from an undergrad spent at the Texas Advanced Computing Center competing in the Student Cluster Competition and a student career working with legitimate open source / open science heroes like Matt Vaughn, Lincoln Stein, Richard Durbin, Benedict Paten, and Erik Garrison, I practically begged for a job (though I did threaten to “go to Google” at my interview, as they had a nascent health team that had recently released DeepVariant; I am extremely, extremely glad they didn’t call my bluff).

And what a job it was. Over my time there I got to help:

• release much more open source bioinformatics code (GenomeWorks, VariantWorks, and other tools for genome assembly and analysis)

• Develop new state-of-the-art tools for VCF merging, annotation, and filtering in Parabricks

• See Parabricks made free for research and development, integrated into the Broad Institute’s Terra (the platform I used and contributed tools to for my PhD work), and installed with almost every collaborator I had ever worked with (or heard of)

• set a Guinness world record for fastest genomic diagnosis using Parabricks

• Install Parabricks on NIH’s Biowulf and at Sanger, my previous home clusters

• Create Parabricks workflows with Gary Burnett, where we crammed accelerated alignment and variant calling into both WDL and NextFlow (and we were thrilled to have people in the open source community come find us to help!)

• Work with National Cancer Institute to accelerate the very pipelines I’d built as a graduate student (and save them a ton of money on the cloud)

• Supervise more than a half dozen all-star interns and junior employees, and see all of them get promoted, go to incredible positions elsewhere, or return to NVIDIA as killer FTEs years later

• Personally interview more than 120 people to grow the new BioNeMo team to dozens

• Have our work mentioned in four of five GTC keynotes (and get to go to two GTCs in person after years of remote ones)

• Lead development of fq2bam, UniversalVariantCalling, AlphaFold2, AlphaFold2-Multimer, MSA Search, and Boltz2 NIMs and the Generative Virtual Screening and Protein Binder Design Blueprints with Ohad Mosafi

• Lead productization of BioNeMo SCDL, the Single Cell Data Loader, which remains state of the art for scalability

• Lead panels on cloud computing and scaling genomic analysis to hundreds of thousands of samples

• Contribute to many cool internal projects that never saw the light of day but which I am so proud of

• Collaborate and publish with so many friends from every stage of my career

• See NVIDIA become the most valuable company on earth, and one driving the future of accelerated computing and AI in every field.

The company itself is incredibly caring, top-down. It helped us through the pandemic, it gave us raises when the field was seeing layoffs, it told us to remain calm when the stock whipsawed 10% in a single day (more than once). During every crisis internal or external, leadership prioritized us and told us to just remain steadfast in our mission. I learned a lot about leadership at NVIDIA, and I will never forget it and will always have immense gratitude. It is truly special.

The question on everyone’s mind (and tongue) has been “Why in the world would you leave?”

To be continued, but, for now, I’ll say that I’m going back to my roots, taking some time off to refuel, and enjoying the memories of all the good times and good friends. Accelerated computing, AI, and life sciences aren’t going anywhere in your life or mine.