Any Data Scientist Mums?

[bluebunny1]

I was wondering if anyone here works as a Data Scientist and would be able to explain in simple terms what their job is like?

The reason I am asking is that I am thinking of applying to an online Masters in Data Science and it would be a career change for me.

What kind of tasks do you do day to day? What challenges do you face? Is it stressful? Do you have many deadlines?

Thanks so much in advance!

DH is a data scientist but I'm not. (I work as a software dev so have a good understanding of their work). His team are all physicists and mathematicians. I hope you like maths. They use mathematical models and AI techniques to solve problems. They are using a lot of ChatGPT currently. They write code in Python and interface using API and also sometimes direct database access. (For example, they access ChatGPT using API, and also they have rebuilt open source large language models on AWS as part of their work). It is not excel or clicking on mouse but a lot of coding, just not to the rigour of a application development team.

Day to day work is actually really similar to software development. Their priorities are set by department wise objectives which are over their paygrades. They get tickets in Jira, but use a much longer sprint then typical in software teams. Each member work on their tickets independently, but together the tickets are supposed to create something bigger (called sprint goal). They commit their changes into Git which is automatically tested and then deployed to AWS. He mentioned a lot of various Python maths libraries like tensorflow, jupyter notebook. They do demos at the end of a sprint.

I don't think the work is stressful. But it's a very tech oriented job and you have to love it or you'll struggle to keep up with the changes. As for deadlines, similar to software teams, you usually have daily scrums so you need to tell everyone what you have completed daily. Your changes are all peer reviewed too.

Oh one thing. If you are slow, you are likely to block other people's work because the team works on interelated tasks. I have never had a problem with this but I know it's awful for those who can't keep up.

Thanks so much SpringBunnies, that's very useful. I do like maths, but don't have a STEM degree (although I did Statistics / SPSS as part of my Cambridge degree). My previous role was in investment banking, but it mainly involved excel, rather than databases. I used to like algebra / linear algebra at school and I gather Python is not too difficult to learn with a bit of time and effort. The masters in Data Science is in itself 2 years + I'm planning to take 1 year for Python.

It is good to hear that the work is quite collaborative and team-based. I know this is a very fast paced and changing industry. Do you know from your husband if he is expected to learn a lot in his spare time or if there is a training budget to keep up with technical changes?

A good company should have training time. Time is usually the most expensive element in training. As an example, generative AI is the big thing. His team made a proposal to try it out on some of the problems they got. They got approval and was able to learn it as part of their work. Hope this make sense?

My team also has a day every two weeks for self learning. This is in addition learning new framework/tech that are directly related to our assigned tickets.

If your previous role is in investment banking, look for opportunties within it. Domain knowledge is invaluable.

Makes perfect sense, thank you!

SpringBunnies' DH's experience is valid for team structured like a large IT department but that's not the only model. Data scientists within quasi-actuarial teams can have a much less structured role in smaller teams. It's an evolving profession so there's a lot of variation - ask lots of questions about how any job you're looking at works.

You should have a huge advantage leveraging your existing knowledge combined with a data science qualification if you can find jobs in that area.

Thanks MsWarrensProfession, can I ask what you mean by the "quasi-actuarial" teams? I understand that a data scientist role in a non-tech company (e.g. energy company, bank) would be different to somewhere like Microsoft, but not sure exactly how?

I manage teams of data scientists to deliver our solutions. And your biggest asset will be your previous business experience/domain knowledge. The ones who struggle are those who approach it from an academic background - and don't understand the most efficient/accurate/user friendly way to develop assumptions, code and solve for hypotheses. The more real world experience you have in the domain, the less hand holding you need from business stakeholders. Writing code isn't the challenge, it's writing code that's not bulky, complex, hard to maintain or needs constant updating - and not many data scientists can do this well.

There are tight deadlines. Where I work we operate in sprints and every task you do would be estimated for effort and managed daily. As someone upthread mentioned if you can't operate at pace, you may slow others down and that's pressure. There is also a lot of regression testing and documenting that's required i.e ensuring you don't break someone else's work while you're bug fixing or developing - but you'll have support around this from your manager or scrum master or solution architect. There's daily progress update calls so you can't ever slack off. Most important is specialising in an industry/domain as you'll find it easier to grow and progress.

There's no real difference between working for a tech company or a non tech company other than - with a tech company you could work more with clients as a consultant rather than purely in house. Because tech companies need to sell their solutions to clients for revenue. So that will be more pressure than working in house.

Data scientist skills

The role can involve
Sourcing good quality data
Persuading the customer that you need good quality data to get good quality results
So you need good people skills - to get the data , to understand the data ( domain knowledge is vital ) , to repair the data - you may need to organise data gathering which could be finding databases or it could be trawling the internet
The analytics - that's the east bit
Interpret the results - what does it mean for the business
Presenting the results in a useful way - is this a report or a regular spreadsheet or do you need to set up a web site to allow users to run queries and select charts etc

dayofmeh, thanks very much for this useful info. Can I ask if your team are generally happy and find that they have a good work / life balance?

midgemadgemodge, this is an interesting point. Are you ever in a position, as a data scientist, to advise on how information gathering can be improved (if the data is not good quality). For example, improved training for frontline employees?

Also I don't think a masters in data science is necessary. The people we hire and I have hired in my previous roles are either MSc or phds in a relevant subject (supply chain, finance, AI, biomedicine etc - something quant) or people from coding boot camps.

I'm not sure the MSc will teach you anything valuable that you wouldn't learn on the job anyway? It would be like doing an MSc in Investment Banking.

Yip we are currently doing training for employees !

Dayofmeh, I kind of do have an MSc in investment banking (I'm a CFA charter holder), but I think the main reason I was looking at a masters in data science is because my main degree is in Social Sciences (admittedly with Statistics in it), so I would feel more comfortable if I can properly refresh my maths, statistics and learn to code. Otherwise I won't feel confident that I can do a good job and don't want to set up myself to fail.

Amazing!

We've not lost any people so far, so I would say they are generally content. Pay scales are probs the biggest bone of contention as this is industry dependent. So in retail and consumer good where I work - the projects and problems are big and very exciting (as they touch so many consumers and you have access to a lot of data) and work life balance is good but it's not as well paid as if you worked for a tech giant or in finance where you'd have worse work life balance. So that's a decision you'll have to take as well - what industry.

Also if you work in a more client focused role work life balance is similar to management consulting I.e not good. So if your company needs client revenue to fund and develop its data science arm, getting that revenue will also be a problem you face into.

Size and maturity of the product/tech/data functions is important too. The more established it is, the more support and easier to get access to business teams, good date, automation, career progression. A smaller or less mature organisation you'll have more autonomy but less support and more pressure.

Whatever you do don't join a company where they think a data analyst/business analyst is the same as a data scientist! They're very very different.

Good point re client-facing roles, I think I would try to avoid those as simply haven't got time for 20 hour days any more (with two kids). Point taken re data analyst distinction!!

Fair enough. I would say its most important to understand good statistical analysis techniques and coding. If the course covers it, you'll be well set up. Look into coding bootcamps as well.

Thanks for this practical advice, coding bootcamp is definitely on the to-do list!

@dayofmeh would you mind if I sent you a quick PM?

I'm a data scientist within a specialist sub-field where I already had domain knowledge. I reskilled via apprenticeships. I love the job, but I love both coding and problem-solving, and selling my work/influencing. Many people struggle to do both and as PP said, in practice an understanding of techniques doesn't really go that far without a grounding in domain knowledge to apply it. Even a perfect solution just sits on the shelf unless you can get out there and tell people why they should adopt it.

As a job, I find it stimulating, flexible, well-paid, and fun. It does move fast and it's collaborative. You have to be analytically minded, independent, and very good at teaching yourself stuff - often I have to upskill myself on a new technique, and the field evolved very fast.

Ps. I wouldn't have got on my MSc Data Science programme without being able to demonstrate maths and coding skills in Python. My course was/is mathematically and statistically complex.

When recruiting I look for enthusiasm for the domain , a logical mathematical brain, a flexible rounded person with reasonable interpersonal skills

I assume I can train the coding skills

Not recruiting at the moment however

Thank you, can I ask which MSc programme you did? Also you mentioned that you re-trained via an apprenticeship, is that in addition to your MSc?