How I switched careers from software engineer to AI policy operations

post by Lucie Philippon (lucie-philippon) · 2025-04-13T06:37:33.507Z · LW · GW · 1 comments

Contents

  The job: International AI policy event organizer
  How I got the job
  My theory as to what led to me being hired and how it can be reproduced by others
  My experience at this job and where I stand now
None
1 comment

Thanks to Linda Linsefors [LW · GW] for encouraging me to write my story. Although it might not generalize to anyone else, I hope it will help some AI safety newcomers get a clearer picture of how to get into the field.

In Summer 2023, I thought that all jobs in AI safety were super competitive and needed qualifications I didn’t have. I was expecting to need a few years to build skills and connections before getting any significantly impactful position. Surprisingly to me, it only took me 3 months after leaving my software engineering job before I landed a job in AI policy. I now think that it’s actually somewhat easy to get into AI policy without any prior experience if you’re agentic and accept some tradeoffs.

The job: International AI policy event organizer

From January 2024 to February 2025, I worked on organizing high level international AI policy events:

• Responsible Adoption of General Purpose AI, a2-day seminar for the Pacific Economic Cooperation Council run by Nicolas Miailhe, which got policymakers in the Pacific informed about AI risk and produced recommendations for the AI Action Summit and for APEC
• AI Safety Connect, a side event of the AI Action Summit run by Cyrus Hodes. As the summit organizers basically removed safety from the agenda, our event was the main place where the experts and policymakers gathered in Paris for the summit could go to hear about and discuss AI Safety concerns.

For both of those jobs, an AI policy expert was delegating to me the task of achieving their vision for the event. Basically, my role was to do all the event production so they could focus on getting the right people to come and the right conversations to happen.

The day-to-day job was mainly split into:

1. Writing emails: sending invitations; managing attendees, speakers and stakeholders; managing vendors
2. Writing Google Docs:
   1. Research-like documents (e.g. who to invite, which venue to use, which theory of change to follow, what to schedule and when)
   2. Procedural documents to run the event (e.g. project tracking, checklists, any doc used to coordinate with the rest of my team)
   3. External facing documents (e.g. event schedule, logistical information for attendees, session plans for speakers, grant application)
3. Project management: Keeping track of all the tasks to be done by the date of the event, making sure everything is on track, deciding what to work on, meeting with other colleagues, communicating on my work on slack, getting the event organizer to approve decisions or give me feedback to build the event they wanted.

All in all, it was a pretty typical operation job. My AI Safety knowledge was mainly useful during the initial planning phase when building the schedule and deciding who to invite, and after that it was mostly grinding through the tasks to get everything ready in time.

How I got the job

In Summer 2023, I got to the end of my software engineering studies. Realizing that I did not have “but I need to keep my job to fulfill the requirements to get my degree!” as an excuse anymore, I decided that it was the right time to leave and search for a way to work on AI Safety.

I did not have much of a plan then, but I was in a favorable position:

• I was living in a group house with EAs, two of whom worked on AI safety, whom I could ask for advice
• I had volunteered at CeSIA, a French AI Safety org in the past, so I had opportunities to skill up there
• I had read a lot of AI safety content and attended multiple EAG conferences, so I had a good grasp of the various parts of the field, the organizations working in it, and the types of roles I could get

However, I did not have many legible AI safety credentials, so my first priority was getting more legible experience. My time was split between:

• Applying to any AI Safety opportunity I found that seemed interesting and possibly reachable (training programs, fellowships, entry level jobs). As I filled out more and more applications, it became faster, as I got better at writing, describing my skills, and reused past application’s answers. Also, application work tests were actually very good training
• Volunteering at CeSIA. I did lots of odd jobs like running events, proofreading documents, facilitating meetings, building processes to check up on org members regularly, writing strategy proposals. I mostly did whatever interested me at the time, and looked specifically for places where everyone in the org was dropping the ball [LW · GW].

In December 2023, my friend Charbel, executive director of CeSIA, got contacted by Nicolas Miailhe, who was searching for someone to run a policy event for him. Despite seeing the job ad being shared in our Slack, I did not pay attention because it did not seem to fit my situation: it was an internship, and I was not a student, and it required event planning experience. That’s where having volunteered at CeSIA proved useful, as Charbel went on to contact me personally to ask if I would be interested in the position, as he knew I was available and looking for impactful AI Safety work.

After an initial round of email, Nicolas asked me to list my relevant experiences and rate myself on relevant skills. For this, I benefited a lot from talking with friends who could give me some outside view on how skilled I was at “autonomy”, or “research”. It was also useful to discuss with others whether this was a good opportunity for me and whether it was impactful.

And then I got the job!

My theory as to what led to me being hired and how it can be reproduced by others

Here is the list of factors I think led to me being hired in this first AI policy job, ordered by importance:

1. Getting experience and referrals by volunteering: Charbel recommending me directly to Nicolas, which he did because of my volunteer work at CeSIA. Nicolas told me that was the main reason he accepted my application.
2. Getting involved and trusted in my local AI Safety community: Without two years of involvement in the French EA and AI Safety communities, I would probably not have the opportunity to volunteer to CeSIA, I would not have heard of the job offer, and I would not have had people to advise me on my career transition.
3. Focusing on jobs I’m uniquely suited for: Nicolas needing someone ASAP who spoke French and knew about AI safety, which probably limited his options to just me.
4. Having financial slack and experience with unconventional employment: Thanks to savings from my previous job and unemployment benefits, I was able to accept the job even though the salary was low. In the end, I ended up receiving a grant, but the financial freedom allowed me to take the offer even with uncertainty over receiving any grant. It also helped that I had experience freelancing, and so Nicolas could hire me as a contractor easily.

My experience at this job and where I stand now

Overall, starting a 9 months long project of organizing a high level policy event with little experience and supervision ended up as well as you could hope: the event did happen and went well, but preparing it was freaking hard.

If I could give past me one piece of advice, it would be to condition accepting this job on getting a specific amount of mentoring per week. Skilling up in a new field on a deadline is hard enough; no need to do it without any guidance from someone more experienced.

I ended up getting quite good at event planning, somewhat better at policy research, and far better at autonomously managing a difficult project without getting stuck in loops of stress and avoidance.

It was the hardest job I had done in my life, and I somehow managed to rise to the challenge and leave better than I came in, and for that I’m grateful to have gotten it despite the difficult periods.

---

For any AI safety newcomer who would like to know more or get advice on their career strategy, feel free to DM me on LessWrong. I’d be happy to have a call 🙂

I’m now taking a break and visiting the Bay to build connections and hopefully find a new project which build upon the skills I built. If you know of such opportunities, send them my way!

1 comments

Comments sorted by top scores.