I’m afraid that I do not write this article from the standpoint of having cracked the problem! But I do have some thoughts, which I’m jotting down here – notes to my future self when impostor syndrome rears its ugly head, as it surely will!

Impostor syndrome is a psychological phenomenon where, despite often overwhelming evidence to the contrary, one doubts one’s abilities and competence. 3 weeks ago I graduated from the University of London, receiving an MSc in Data Science with distinction. You would think this would be the final nail in the coffin of my impostor syndrome, and yet disappointingly the monster is still out there!

Anne-Laure Le Cunff of Ness Labs outlines the five types of perfectionist – which to me seem to have some overlaps. I identify most with the superwoman type (who “work harder than everyone else in an attempt to hide what they believe is a lower level of competence”) but also relate to the perfectionist type (who “set excessively high goals for themselves”) and to a degree the soloist type (who “see asking for help as a weakness, and try to do everything on their own”).

Before switching careers to become a data scientist I had been an SAP consultant for many years. At the point where I made the switch I was considered an expert in my field, I was at the top of the pile. And indeed that was part of the problem for me: the work had come to feel rather routine. Starting out in data science put me at the bottom of the pile suddenly! That was uncomfortable and anxiety-provoking.

So what helps?

Commit to a process, not a goal

James Clear’s advice to commit to a process rather than a goal helped me a lot in the beginning, and continues to do so. It prevents me from feeling overwhelmed by those afore-mentioned excessively high goals. For example, instead of telling myself “I need to learn to code”, I will rather tell myself “Today I will work through one lesson on my coding course, and then I will find a way to apply that lesson in a new setting so I can understand it outside of the lesson context”. If it happens that I achieve this and I still have time left over I can celebrate my achievement and go a little further. But if I struggle a bit it’s perfectly okay because I’ve committed to immersing myself in just this one thing and really getting to grips with it. The focus is on the quality of the process, not on reaching the finish line.

Look back, as well as forward

Obliquely inspired by a section in Slater’s book “Jung vs Borg” I have come to appreciate the dangers of living too much in the future. Of course we are moving towards the future and we have our aims and objectives, hopes and dreams. But if that is the space we dwell in exclusively we miss out on the richness of reflecting back in time, and the emotional support that can give us when in the grip of impostor syndrome. For instance, if I take some time to look back on the past year I can of course recall all the fears associated with whether or not I would successfully complete my final MSc project and thesis. BUT I can also appreciate (dare I say admire even!) how I kept calm, reached out to friends, family and colleagues for both emotional and academic support, found the extra time I needed by negotiating some unpaid leave, improved my focus by using a co-working space. I can also appreciate that, despite genuinely being afraid that I would not cook up a satisfactory algorithm for blending all the NLP components I needed to in order to build a knowledge graph, I did in the end crack it and the final result was robust, modular code that ran well even when challenged with all the variations found in >2800 news articles.

Failure is a learning opportunity

Like anyone I find failure unpleasant, uncomfortable and undermining. BUT failing has a wonderful silver lining: it can be one of the best ways to learn deeply. I usually can’t help wallowing in the misery of failure for a while, but what clears that misery away is the point where I start saying to myself “I wonder what I could have done differently?”, “I wonder if I tried a different approach?”, “Perhaps this problem needs a bit more research…”, “I wonder if someone else might have insights for me on how to avoid this in future?”. And so my thoughts turn in a more positive direction. Spending these years in an academic environment has also given me a new appreciation for the depth and breadth of research that is possible. I’ve come to realise that when I think I’ve reached the end of my options, there are usually still a dozen out there waiting to be uncovered.

Internalise that compliment

We have an annual peer review process at my company. Every year my colleagues write the most astonishing things about me! Apparently they value my contribution to the team, my work ethic, my technical skills, even my personality… And every year I think to myself “Thank goodness they haven’t realised how useless I am yet”. This is not productive! I am trying to coach myself into accepting compliments. I don’t think it’s unusual in society for us to reflexively deflect compliments – but in the spirit of looking back, it may actually be more useful to reflect on whether I can remember a time where I produced a good piece or work, facilitated team communication, or offered to stay late and finish up a task. In that way I can try to really own the nice things people say.

Reach out, they won’t bite!

Impostor syndrome can keep us isolated. But starting at the bottom of the pile in a new field I have had to learn to reach out for help. I think there is a balance: some struggle is good for learning and growth, but there also comes a time when one gets unnecessarily bogged down and a helping hand is useful. I must say that, in my experience, the data science community has been nothing but friendly, interested, engaged and level-headed. I think we are the kind of people that just enjoy seeing something new, chewing on a difficult problem, and trying to think about a thing from multiple angles. Although the prospect is still sometimes daunting, I have consistently found that collaborating on tasks is not only rewarding, but produces better results, and usually enriches my knowledge and understanding – and I think that of the people I work with too.

I’ve come to realise that in data science we are all continually learning. Almost everything we get asked to do or build will involve learning something we didn’t know how to do before. This is what keeps it interesting each day! Rather than fearing my ignorance on some topic will be exposed, I aim to use that ignorance as a springboard to further learning ♥︎.

Post-script: AI tools and impostor syndrome

06 Jul 2025 ~ In the past months I’ve grappled with the use of AI tools, which can induce their own kind of impostor syndrome: “I should have been able to write that function myself.”, “Am I becoming too stupid to code on my own?” and the like… I have been trying to remind myself that struggle is good as it means I’m really learning. Recently I was delighted to read an MIT research paper recommended by a former UoL colleage @FabioDePonte entitled Your brain on ChatGPT (2025) which corroborated this standpoint. Participants were divided into 3 groups (LLM usage allowed, search engine usage allowed, and brain-only usage allowed) and challenged to produce essays. “EEG revealed significant differences in brain connectivity: Brain-only participants exhibited the strongest, most distributed networks; Search Engine users showed moderate engagement; and LLM users displayed the weakest connectivity. Cognitive activity scaled down in relation to external tool use.” Furthermore, “over four months, LLM users consistently underperformed at neural, linguistic, and behavioral levels.” Strong evidence that spending time in the “struggle zone” is worthwhile and productive: a sign of active learning rather than any inherent personal or intellectual deficiencies!