Harry Gray, Lee Wall, Chris Stamper
Coefficient Giving (CG) funds career transition grants: a few months of support to move into a priority cause area. The three of us used one to move into biosecurity. Harry came from a self-funded tech start-up, with a biostatistics PhD and a forensic science postdoc behind him. Lee came from building CBRNE terrorism risk models for US government clients. Chris came from an immunology postdoc.
Most existing write-ups of mid-career transitions are personal narratives (Jeff Kaufman) or are not specific to biosecurity (Martyn James, Andrés Jiménez Zorilla, Cameron Holmes, Moneer). They are interesting, but hard to act on. The exception is this post from Successif. We have tried to write something practical instead: the questions we were asked, or wish we had asked, answered by each of us in turn. Our routes differ, so read across the answers rather than treating any one as a template.
We have no insight into how CG makes grant decisions. Nothing here is advice on how to get funded.
Answers are labelled HG (Harry Gray), LW (Lee Wall) and CS (Chris Stamper).
What was your work before the transition?
HG: I worked at a growing self-funded tech start up. I joined after the co-founders so I did anything needed to grow the company and this developed a generalist skillset. I also had specific expertise from a PhD in biostatistics and a postdoc in communicating uncertainty in forensic science.
LW: I worked as a data scientist at Battelle making models for quantitative risk assessment of CBRNE (chemical, biological, radiological, nuclear, and explosive) terrorism for U.S. federal government clients. This involved a lot of darting in and out of SCIFs, navigating labyrinths of legacy C++ and C# code, reasoning about threat models and pathways to harm, and juggling a lot of context between projects and models. I sometimes had to step in as an “expert” on a model I had worked on for 2 months a year prior because the people who had made the model had long moved on to greener pastures.
CS: Lifelong academic scientists, more or less. I had finished my PhD at the University of Chicago in Immunology and Computational Biology, focused on antibody responses to influenza and SARS-CoV-2 and trying to improve vaccines to both. I moved to Sweden and did a postdoc in tissue immunology, applying high throughput, computational, and ML techniques to understand human gut, innate immunogenetic networks. I had been consulting with various biosecurity topics on the side during my postdoc.
Why did you decide to transition?
HG: I felt a strong pull to work on impactful problems. I had been a 10% pledger for a few years but I felt I could do more. Over time my thoughts on how I should spend my career developed and gradually became misaligned with my work.
LW: Confluence of factors. I was starting to feel a real sense of urgency around doing something impactful with my life and around the hazards of technological progress, which has long felt to me like an unowned problem of civilizational scale. Working on WMD terrorism models had deepened this mindset, but it (1) still didn’t feel right to me to work on these problems primarily from a national security angle, and (2) I didn’t think these models were actually moving the needle much on reducing these risks. Somewhere along the way, I was reading a lot of philosophy online, encountering things like Nick Bostrom’s Vulnerable World Hypothesis, and eventually found out about the EA community. I had no idea that there was this little world of people orienting their careers toward problems with global, existential stakes, and immediately knew I wanted to jump in.
CS: During the four years of my postdoc I had been consulting on the side with different orgs working on topics related to biosecurity. This included Alvea (flu vaccines), SecureBio (VCT and ABD-Bench), and MBDF as a ‘scrutinizer’ on the Technical Report on Mirror Life. That work and some friends in the EA space drew me more into the circle of caring about existential and catastrophic risks so I decided I wanted to work full-time on researching and understanding how we could prevent them or safeguard ourselves against them.
How did you decide the time was right to transition?
HG: The start up was/is very successful and we had added career professionals across key functions. The timing was right to leave because no critical functions were left uncovered.
LW: I was itching to leave by the time I did, and several folks at CG were nudging me to quit my job. My own risk-aversion kept me from doing it for longer than it should have, so I think it was helpful that someone outside my head validated the decision I knew I wanted to make.
CS: Karolinska Institutet, where I was a postdoc, has a hard four year cap on postdocs. I was offered to continue on as an Assistant Professor so it created a natural decision point. I decided to take the leap and turned down the Asst Professorship and sent the remainder of several grants I had been awarded, some of which had years of funding left, back to the Swedish organizations that had given them to me.
How did you prepare for the transition?
HG: I communicated openly in advance with my work colleagues and so we began a process to transition functions away from me within the business. I also built up an emergency fund of money specifically to sustain myself throughout the transition. I followed the 80,000 Hours guidance and planned cheap tests across a range of interest areas. I did not want to apply for the grant immediately as I wanted some time to reorient and explore using my own rules first.
LW: I had several projects that I’d been brought on for which were still in their first 25% of life, so I had some discussions with project managers prior to applying for the grant where I explained the situation – which to most in the normal professional world, is really weird – and came up with a plan for what I would hand off and what I would finish before leaving. It ended up working out pretty well, as by the time I actually left, some projects were in funding limbo, my timecard would’ve had a bunch of weird holes if I worked another two weeks, and my (very supportive!) manager helped arrange for me to leave just two days after receiving the news I’d gotten the grant.
CS: Well, I had to leave Karolinska either way so my PI and I were already at work on a handover plan for all my data and ongoing projects to other trainees in the lab. I also had been in contact, before ever applying for the grant, with MBDF. I had worked as a notetaker at their Paris Conference and they had a project that needed an immunologist, and would especially benefit from an immunologist with B cell expertise. So my grant was structured around the first 3 months of it being at MBDF to work on that project. I also started talking to as many people as I could in the biosecurity space to figure out where and what I wanted to do next. Just before applying I was able to attend EAG London and have a lot of valuable conversations there and made many valuable connections I would later use to find my grant ‘rotations.’
Why did you choose biosecurity?
HG: I had not chosen biosecurity when I quit my job. I was more interested in AI safety at the time. I ruled out a few areas like AI safety governance and red-teaming through cheap tests, but I didn’t find an obvious fit. I had ruled out biosecurity tentatively as I did not want to revisit my former PhD life. 80,000 Hours Careers advising convinced me to reopen the door to biosecurity.
LW: I’m also still quite interested in problems beyond biosecurity, but I generally had the sense that the theories of change for AI safety work are far murkier than “build a physical system to detect engineered pathogens in wastewater” or “get PPE in place for every essential worker”. After spending three years in the world of government contracts and national security, I was sick of murky theories of change and ill-defined threat models, so I thought the most reasonable move was to do something clear and concrete but still existentially important. Two additional reasons are that I felt biosecurity was more urgent than other problems, and that most plausible pathways to extinction or civilizational collapse routed through bio-ish things anyways.
CS: I have been intellectually interested in pandemics and was working towards their prevention long before I found EA. A research paper about survivors of 1918 influenza still having viable antibody responses to 2009 Swine Flu (a descendant strain of 1918) is a major reason I decided to study immunology. I spent years working on projects related to universal flu vaccines and was, in some sense, right at the research frontline to SARS-CoV-2 (I helped clone the world’s first non-spike antibodies and some of the world’s earliest spike antibodies too). As I became more interested in catastrophic risks, bio-risks was a natural fit. Once I started the transition I realized the bit of comp and ML background I had also made AIxBio a natural slot for me to try to find a place in.
Why did you apply for the CG grant over alternatives, e.g. keeping your job and upskilling in your free time, applying for a fellowship, applying for another job, saving money to self-fund?
CS: Well, my job had to end unless I did the full on academic route to professor. I had one project lined up already with MBDF and other potential projects with SynX and SecureBio and was being urged to apply to the CDTG, so it made sense to just do it. “Saving money to self fund” as an academic is not really…tenable, I would say.
LW: I think I was trying to do the whole “upskill in your free time and save to self-fund” thing, but this just started to look way less likely to lead to anything than plugging into the right networks and sprinting on projects. I’m also just skeptical that upskilling really goes anywhere in general; if you’re not producing something, I would guess that the outcome of this is that you just forget the skills from disuse pretty fast. Also, this wasn’t mutually exclusive with fellowships for me – I applied for and did a fellowship very soon after I began the grant period. It’s possible that the fellowship alone would have been enough if I was optimizing for landing a different job immediately, but the CG grant was useful for giving me time to build my own views of what’s worth committing to rather than jumping at the first opportunities that present themselves.
HG: I didn’t see any other suitable funding at the time. I worked out that it would take me about twice as long to transition with a part-time contracting gig and so this was a strong motivation to apply for some funding to help cover costs.
How long did the application take and what do you think was strongest about yours? (Note: We do not have any insight into CG grantmaking, so take this as-is and do not use it as advice.)
CS: I did the entire thing on my hour long train commute between my postdoc and home. I think three key things about mine were just all my past relevant experience and skills, past work with EA biosecurity orgs, and having a clear plan for a first thing to do with it.
LW: Not long at all. I knocked it out in <1 hour after EAG NY in October. By that point, several people were nudging me to apply, I’d had some conversations with people who expressed interest in mentoring me for a work rotation if I did the grant, and folks at CG were asking me things like “Quit your job yet?”, so I think the written application had comparatively little to do with it, while having a plan for how I’d use the time and a lot of frontloaded conversations were more important. Could be wrong about that, but I basically think my written application was meh while my value-alignment, quantitative background, prior conversations with people in the field, and plan for how to use the time were strong.
HG: It took me about 10 hours. I wanted to think carefully about my plan and how much to ask for. My mix of experience in a start up and relevant academic work was probably a strong signal.
How long was your grant period and, in hindsight, did you need more or less time?
CS: Nine months! I think it has been a good amount of time. One year could also have been good, and shorter may have worked too but I have been quite happy with the 9 months.
LW: 6 months, with a 2 month pause for the ERA AIxBio Fellowship. Honestly, I probably would’ve given it an extra 2-3 months because (1) I ended up using some of the time to keep thinking about my ERA work instead of immediately starting another rotation, (2) trying to think from first-principles about where I’m best placed and have lots of conversations takes time (one person even suggested I spend the whole rest of the grant just having conversations and building understanding instead of focusing too much on outputs, which you can get through jobs), and (3) if you’re doing 2-month work rotations, 6 months may not be enough for a diverse sampling of kinds of work, organizations, and problems; I find myself wishing I’d done a policy-facing or physical-engineering-flavored project just to test fit. That said, you have to balance exploration time with the particular urgency of biosecurity, so I doubt I would’ve gone past 9 months.
HG: 6 months. This was the right time for me.
How did you use your grant?
CS: To work at different organizations to explore different cause-areas of biosecurity I could work in AND different organization styles (policy, startup, non-profit etc) I fit best in. I also used it to explore different types of work (more or less technical, research managing etc etc) to see where I fit best.
HG: I used it to subsidise living expenses and pay for conference and travel costs. I did many cheap tests to get a sense of fit and tried to meet as many people in the field as possible. I quickly found my way to longer projects in good-fit areas.
How closely did you stick to your original grant plan?
CS: I think pretty close? My original plan was, to be a bit reductive, 3 months with MBDF and then ‘explore other organizations and types of roles’ and that worked out.
HG: Pretty much spot on. I quickly ruled out ops work and pure AI safety to focus on biosecurity.
What was the biggest positive surprise during your grant period?
CS: How much one can upskill in 9 months if you never say no and tackle everything with passion, curiosity, and dedication.
LW: People in biosecurity are generally really happy to make intro’s and talk to you as a new person on the scene, which makes these career transitions much more effective than if it was just money and a send-off. Also, you can get up to context and develop your “taste” really, really quickly. Sometimes it only takes one conversation or an hour of researching a problem to surface a crucial consideration that determines if it’s a priority or not.
HG: The community is small and welcoming. Lots of people were willing to give up their time to talk with me. Those conversations and in-person events were more valuable than any other activity I did in terms of getting up to speed and figuring out what to work on.
What was the hardest thing about your grant period that you didn’t expect?
CS: Sometimes figuring out what to do next and how. How to set up projects or ‘rotations.’ How much freedom I had to do things, etc.
LW: Agree with Chris^. Setting up rotations is sort of weird, and I underestimated how much time could fall out from under you when considering options, spending time thinking about a problem outside of a rotation, or just waiting on busy people. In a 6-month period, a 2-week side quest is nearly 10% of your time! Would recommend either setting firm preferred start/end dates for your rotations and locking them in well ahead of those dates, or budgeting wiggle-room time for decisions, networking, side quests, and potential delays.
HG: My views on what I should work on post-grant changed throughout the grant. This created a lot of uncertainty for me as it wasn’t clear who I could work with or where to seek funding.
What do you think makes a transition into biosecurity different from other career transitions?
LW: I haven’t done any others, so not sure. If I had to guess, biosecurity might have tighter networks across different areas than a larger field, where networks might cluster by subfield (e.g. mechanistic interpretability). Also, the EA-biosecurity community is operating with an unusual sense of urgency, with many tightly scoped problems and not enough people to move them forward. I think it might therefore be more amenable to career transitions, since a spare, highly committed person is less likely to end up floating.
HG: The biosecurity community is small. There are also senior experts who have spent their careers addressing biosecurity issues since WWII. I’d recommend reading and networking more outside of EA-aligned work to learn more biosecurity fundamentals from primary sources early. I like Greg Koblentz’s work.
What would you realistically have done without the grant?
CS: Taken the professorship? Or having turned that down, worked in biotech or something while trying to break into the field.
LW: Worked another 6 months in my last job and kept looking for another exit point. I was basically set on leaving, but the temptation to build more of a cushion would’ve kept me there longer.
HG: I would have done the same but it would have taken me much longer.
What are your recommendations for career transition funders?
HG: An easy win would be to create more structure and accountability by pairing grantees up together to check in regularly. Or otherwise creating a low-overhead way of grantees to connect with each other to share the experience and what’s working with each other.
Photo by Javier Allegue Barros on Unsplash
