The title says it all, I want to transition from actuarial science to data science.

My background: I have a BS and MS in pure mathematics, both with high GPA (3.94) and graduation with honors. My areas of study were highly theoretical (Algebraic Geometry), though I took some statistics courses, computer science, and economics as well. I am savvy with R, Python, Sql, etc. Unlike some pure-mathy types I do communicate well and I do well at translating business questions into quantitative problems.

My motivation: I'm not really motivated by money, as most know actuarial science is already fairly high paying. I think I would enjoy a career in data science more than actuarial science. From the research I have done, the skill set and projects in the data science domain seem more intriguing to me. I have always been intrigued by deep learning, machine learning, etc. but since it wasn't directly relevant to my studies in school or my career I have never built a strong background in these areas.

My Attempts: I have sent out some applications for data science positions but haven't had any luck yet. I am located in the Seattle area and being a tech-y city I imagine there are a lot of competitive applicants for any jobs that get posted. I usually get a "thanks but no thanks" response within a week.

The question: How can I make myself more competitive and at least land some interviews. Since I have only worked in actuarial science (3 years experience), my resume and work history are only actuarial (however my last title within the actuarial domain was actually "Actuarial Data Scientist", we were making web applications for insurance clients that let them view models we built for policyholder behaviour and other things of interest.) Most of the job postings request NLP, deep learning, machine learning experience, to which I have little exposure. I have been self-studying with PyTorch. What types of things can I be self-studying to make my skill set more desirable, and how can I communicate these skills through my resume? It seems uncommon to have a resume section titled "things I have learned in my spare time".

Obviously there are loads of things I could be studying and learning, so I scan job postings to try and narrow down the most commonly requested skills. What would be the best things I could start with, and then how could I incorporate these skills into my resume or cover letter?


Since you have both degrees in pure math, most likely you won't need to do another Masters or similar. For now, focus on making your CV look more like DS/AI/ML/DL. Keep in mind online courses at present do not increase the value of your CV too much. So this is not an easy task, but you have a good start with pytorch. What I'd do if I were you: assuming, you are interested in applying DL in computer vision. Here are some links:

  1. kaggle.com (competitions)
  2. paperswithcode.com
  3. labelled datasets: Pascal VOC, MS COCO,
  4. torchvision library (model zoo) for pytorch (simply import torchvision to get started)

What you should do, is 1) read through some papers on state-of-the-art OS solutions for CV, like Faster R-CNN, Mask R-CNN, 2a) use them in pytorch, they are available in torchvision and/or github, 2b) same with the datasets, 3) hack into them to understand them at a deeper level, luckily they are all opensource, 4) write your own functionality for them, 5) contribute to contests/OS packages, put up your own models on github

You can put all of the above, especially 4) and 5) on your CV. Good luck!


I think your background could be highly interesting for certain paths in data science. If you are only interested in the engineering parts of it, @Alex answer is right for you, but you will compete against people with a lot more applied computer science/engineering background and experience in data science than you have now. However, if you also like the mathematical parts behind the methods, you could consider applying for niche topics where your background will be highly sought after. One such area would be topological data analysis (TDA) which builds upon algebraic geometry and most people (including me) lack the mathematical sophistication to really delve into the subject.

Last year, I was doing a research internship at a very recognized ML research group and during my time, the professor hired a new PhD student which came from a pure math background with a focus on algebraic geometry and who dropped out of a pure math phd. He started to work with him on a TDA project.

However, I think you would have the best chances following this route through searching respective PhD positions, where genuine interest and motivation about the topic and its mathematical foundations (and not only the tech part) combined with your background should be enough to land you such a job. After having a PhD in ML, in which you will probably get a lot of exposure to other topics than topological data analysis alone (and you maybe even change your initial PhD topic), you will have no problems being invited to interviews for interesting data science positions.

I also want to add that there are some companies and applied research institutes which apply TDA and maybe would be open for someone like you. However, one would need to find them and convince them that you are their man even though you have no real experiences, both of those things are not easy.


Not the answer you're looking for? Browse other questions tagged or ask your own question.