My winding, unplanned path from Doctor to AI Startups (and all the steps in between)

In 2020, as the peak from the first wave of COVID subsided in the UK, I worked my final shift as a medical doctor. Since then, I’ve worked as a data scientist, founded a VC-backed start-up, been an independent consultant for 10+ health AI companies and now work at Anterior.

It took me ~3 years to go from “I don’t think I want to be a full-time doctor forever” to getting my first full-time role outside of medicine (as a data scientist) and it’s been ~5 years since then.

During this time, I used “high-speed career sampling” to gradually go from a vague notion that I liked “tech and scalable impact” to a clearer definition of what I love and how I can add value.

I found it helpful to:

• find people 3-5 years ahead of me on career path and ask them lots of questions
• take “little bets” (eg. committing to a project for 2 months - and seeing how I felt at the end)
• view my professional life in phases (ie. don’t be a donkey)

I don’t believe career paths should be linear and I think everybody should find their own way. I’m sharing my path not because I think you should emulate it - but because I hope my experiences can be a helpful datapoint.

Phase 1: Investment Banking, Management Consulting

The first routes I considered out of medicine were perhaps the most obvious at the time: investment banking and managing consulting. Back in 2016/2017 there were a bunch of medics doing this, and books like Management Consulting for Medics with tips on making the leap.

I considered investment banking first, without really understanding what it was, because a friend left medical school straight into LSEG and was telling me how great it was.

I applied for a few jobs and did a few interviews - but ultimately performed poorly because I didn’t actually know much about investment banking beyond a few evenings reading investopedia.

Shortly after, I considered management consulting more seriously. I reached out to a bunch of medics-turned-consultants and read the classic books. I decided it was a solid choice because working at a top firm would open more doors downstream. I still didn’t really know what exactly I wanted to do outside of medicine and this kept the optionality high.

I spent 3-4 months doing case interview practice during quiet medical nights shifts and did a remote online MBA. After doing the case interview practice I realised, however, that I didn’t love it, and it also didn’t come naturally. I’m much better at writing than I am synthesising thoughts while talking. It felt like the latter is an important part of being a good management consultant.

So I made the call to withdraw my applications to McKinsey, BCG, etc before doing the interviews - and to consider an alternative path.

Phase 2: Public Health

I’ve always wanted a “scalable impact” - and bought into the idea that your impact as a doctor is limited by the number of patients you can directly treat. Public health, however, impacts populations and so doing good work here can (in theory) save more lives.

I reached out to my local council and spent some time working with them. I completed a project looking at the impact of air quality interventions on respiratory problems. The work was interesting but I realised two things.

Firstly, that being a doctor wasn’t a concrete value-add in many cases - there were many public health registrars who didn’t have a medical degree and still contributed in much the same way.

Secondly, and more importantly for me, that having an impact is very driven by the political mood of the time. A public health initiative with great data supporting it may be ignored in favour of one that has more political interest at the time. Therefore you had to be patient and work within the shifting political winds.

I concluded that perhaps you can have more impact up-stream: dictating the agenda and policy rather than implementing it. I’m still open to some kind of role in politics - but I’ll leave that for a later phase of my life.

Phase 3: Programming, Data Science, Machine Learning

Having explored options where soft skills were the focus, I wanted to learn something more concrete. Programming felt like a great option.

Initially, I just wanted to make things with code, without really understanding the different components of that. I’d heard about data science and machine learning, but wasn’t too sure where that fit into building apps.

So I started by taking online courses and building projects (more details here).

I came to realise the relevance and value of data science and machine learning in healthcare, so increasingly leant into that. I loved the need to combine maths and code. The introvert in me embraced the fact I could sit in a room and work on projects for weeks and months. In medicine, you see new patients every day. With projects, you can work on the same thing for a sustained period and have built something cool by the end of it.

This was the avenue I enjoyed the most, so my little bets (ie. projects) led to bigger bets: I enrolled in a tough Master’s degree (Data Science and Machine Learning MSc) at University College London then started my first full-time technical job as a data scientist.

Phase 4: Teaching, Content Creation, Courses

I’ve always enjoyed teaching – as a mechanism to consolidate my own understanding, while helping others. I’d taught basic sciences at Cambridge University as well as English as a foreign language at various universities in Japan and South Korea.

So I decided to try make a career out of this. Throughout 2020, I wrote educational content online, made YouTube videos and ran in-person courses.

This was a lot of fun, and I started to see some modest success. I gained guest lecturer positions at UCL and Imperial College London, and my YouTube following grew to ~9,000 followers.

However, I found the overhead of “packaging up ideas” to be time-consuming – and worried about spending more time teaching than actually doing. (Maybe this is easier now.) I had a solid grasp of the concepts I was teaching but felt like I hadn’t truly built something cool with them yet. Which led me to my next phase:

Phase 5: Start-Ups

By the end of 2020, I felt I’d developed a critical base skillset and now wanted to build something. Starting a VC-backed company seemed to be the way to go, so I joined the Entrepreneur First accelerator and created “Billions Health”.

I pivoted through a bunch of ideas for around a year, centred around enabling large scale machine learning using health data, but then ultimately decided to call it a day.

By that point, I had begun to question whether the VC-backed route should be the default choice and also whether you should start with a co-founder. I read Peter Level’s book “Make” and spent 8 months doing (i) independent consulting to build up my savings and (ii) hacking on my own potential business ideas.

I was determined to keep going until something popped off and was convinced I’d get there eventually. But then a friend reached out with a great opportunity: LLMs were popping off, he’d found a great early use case in healthcare and was raising some initial investment. Given how strongly this aligned with my background and interests, I decided to put my own thing on pause and give it a shot. This later became Anterior - and it’s where I work today.

Reflections

I’m very happy with how things played out, and where I am now, although I never really planned any of this out. I just followed curiosity and built skills I hoped would be valuable. I learnt to program, for example, without a concrete plan of what I’d do with it - or even being sure there was a job that existed for my combination of skills. (I’m not saying I would recommend this!)

I do still have the sensation that I’ve not progressed quickly enough. All of this has taken years. Perhaps there was a more direct route I could have followed to achieve similar ends. But on the flip side, I’ve built a pretty esoteric set of skills which I hope will enable a career path that, although unconventional, is deeply enjoyable and adapted to my interests. In that regard, I feel like I’m just getting started.