Do Your Life's Work

Today I am excited to announce the next phase of my career (after some lovely time off and a lot of work behind the scenes).

If you’ve visited the About page of my website, you’ve seen my three career mission statements¹:

• Make cancer genome analysis free, instant, and accessible to every human being
• Show it is possible to build AI and computing systems that are good for people and planet.
• Ensure we understand and remember the genomic impacts of nuclear testing and disaster.

The first of these started to form when I was an undergrad; the other two became increasingly important to me as a graduate student and as I’ve watched the AI data center explosion take hold over the last four years. They have helped me to make many small and large decisions. Many of my loved ones have been impacted by cancer in their lifetimes. Many also lived close to their environment, whether on farms or as lovers of the outdoors. All of their experiences, my own experiences, and the experiences of many, many others I have met along the way continue to help me grow.

I view these three career goals as extremely related. Cancer is a disease of the genome, with mutations accumulating from both our genetics and our environment (ionizing radiation being one of the strongest “environmental” carcinogens²). A better understanding of each of our unique genetic and environmental histories can provide us with more personalized cancer care and better outcomes. Such personalized medicine requires data (both population data and individual data), and these datasets are huge - we must use computers and algorithms (fast ones) to analyze all this data at reasonable time and cost. We must do so in a way that is sustainable. We saw in great detail with COVID-19 (and every day with climate change) that planetary health is human health. Providing this sustainability constraint, in my mind, encourages innovation, rather than stifles it. I am convinced at this point we’ll see near-instant and near-free cancer genome reports (from blood draw to diagnosis) by the end of the decade as sequencing continues to improve³.

To business - A Pair of New Ecosystem Companies Link to heading

EarthFrame Link to heading

I am now a Cofounder and acting President of EarthFrame, a Public Benefit Corporation that makes building blocks for empowering communities in the data economy and helping humanity sustainably steward what we think is Earth’s most precious data. You can read more over on the website. Here, I’ll just say I am grateful to be working with great friends on interesting, challenging problems that require our best work. I am incredibly proud - we have been stunningly productive, but more than that, we have each spoken up for our values, listened to each other, and built some of the most futuristic, innovative physical product / art I’ve made during my career. My cofounder Yousuf called it the “most fun and most caring place he’s ever worked” (which he admittedly has helped create). That’s an accolade that keeps me going to sleep at night with a smile.

Phileal, LLC Link to heading

I am also working on Phileal, LLC, which builds and integrates accelerated genomic software to realize the goal of instant, free, and accessible (pan)genomics in cancer and non-cancer settings. In addition to optimized workflows for alignment and variant calling powered by NVIDIA’s GPU-accelerated Parabricks suite, I have developed some new libraries and tools for accelerating other parts of germline and somatic analysis such as mutational signatures, pangenome construction, etc. More of these will be formally announced over the coming months.

I have been doing a limited amount of custom consulting work as well through Phileal, as well as some partnership with other companies. I’ve been doing some form of accelerated bioinformatics for 10+ years now, so this is all very familiar - what’s exciting is where I’m seeing Phileal’s work being applied and at what scale. More mysteries for you to await resolution to. For now, the best way to ask what Phileal is up to is to reach out on my personal LinkedIn page and ask to chat about it - I would love to hear from you.

Why two companies? Link to heading

There are many synergies between the companies, but keeping them separate makes the most sense for IP and organizational reasons at the moment. It is possible - and awesome - to buy combinations of Phileal software and EarthFrame products. EarthFrame supplies any hardware bought through Phileal (though Phileal software runs just about anywhere). And EarthFrame, while it builds hardware, also supplies software that runs across ecosystem partner hardware.

Wait - I thought you were taking time off? What have you been doing outside the office? Link to heading

Great question! After a move back to Texas, a burst of business formation and getting the ball rolling on our roadmap, I have gotten to take lots of vacation (which was very much my norm at the former company - I made good use of that unlimited PTO when I could).

In October I taught Cancer Genome Analysis in Khon Kaen, Thailand with the Wellcome Advanced Courses team. These courses are always some of my favorite weeks of the year and I am grateful to have gotten to teach both NGS Bioinformatics and Cancer Genome Analysis several times over the past six years. I snuck out for some vacation in Hong Kong after.

In November and December, I got to enjoy the holidays with family and again got plenty of vacation.

In January, I was nose-to-the-grindstone working. The lead up to releases, presentations, and announcements is always the most chaotic.

In February, I headed to Maohi Nui (French Polynesia), where I am supporting Dr. Vehia Wheeler on a project to build local capacity for measuring and analyzing radiation data using Radiacode scintillation counters, which are super easy to work with and can be had for under $500. Maohi Nui experienced widespread fallout from 193 nuclear bomb detonations on Muroroa and Fangataufa during the French nuclear weapons development program. This work complements what Prof. Keolu Fox, myself, and our team have called “The Polynesian Health Futures and Genomic Inheritance Project,” which aims to create local capacity for genomics and precision health in Maohi Nui.

Spring conference season runs March-May for me - hope to see lots of y’all at GTC!

I’ve kept busy, to say the least, though working with your friends on a beautiful island makes the work a lot more pleasant.

So… why leave Nvidia right then? Link to heading

This sounds bizarre to say, but a big part of the decision to leave NVIDIA was driven by productivity gains from AI. The difference over time was striking - in just 18 months the tooling I was using went from being not good enough to pass a very basic interview question to legitimately producing senior FTE-level work in seconds⁴. I found I was able to execute on my ideas and designs much faster, and I saw a clear path to completing those career mission statements⁵. At around the same time, the proto-EarthFrame team - Keolu, Yousuf, and I - got interest from investors to go do … well, pretty much whatever we wanted it seemed. Which in the end we didn’t take them up on, but that’s a story for the data archives.

I made a promise to myself when I joined the big company that I would stay at least four years and, after that, only leave in pursuit of my career mission, if an opportunity arose. That moment came in October, and I am grateful for the confluence of factors that made it possible. Will this always be what I work on? Surely not - EarthFrame and Phileal’s roadmaps are ambitious, but I expect the businesses to solve the problems they are designed to solve.

I will always bleed a little green. Nvidia in many ways felt like another four-year degree, in that I learned a lot and I am both proud to be an alum and a bit sad to have left (for now). But I didn’t go too far, I now get to partner with even more of the ecosystem, and I get to go back to being deep as can be in the hardware and cancer genomics worlds. I’ll be back at all my usual conferences this year, so find me at GTC or ASHG and say hello. And if you know someone who needs hardware or software for really fast, sovereign, sustainable AI / bioinformatics / genomics, please send them my way.

Gratefully, Eric