This blog post is intended for all those who recently got out of University or finished some sort of online course after deciding to change their career and are starting their first job search for Data Science positions.
Having been in that position myself after leaving UCL after my PhD and ending up in a job I left after two weeks, I want to share my experiences and some tips, so you don’t end up making the same mistakes as me.
Disclaimer: there are more “general” red flags, and spotting them comes with experience and can depend on your own preferences (what one person would see as a no-go might be fine with another), so this post is specifically targeted to those applying for their first data science job.
You're invited! Join SheCanCode's Women in Tech Community
Find a supportive network, opportunities, jobs & much more, so you can excel in your tech career.
0. Cliche driven ( “AI focused”, “data driven”, “insight generation”)
Let’s start with the cliches. If you encounter any of the following lingo in the job spec or during the initial interview, I would not even consider applying for it/pursuing it after the initial call.
Just like you are selected based on your CV, you should use the job description to select which companies to apply to (and where not to apply to). If it is too vague, lists unrealistic requirements, uses buzzwords such as Big Data or AI without a real reason, don’t bother. Chances are someone decided they needed to get a Data Scientist (because that’s what the cool companies do these days) but they have no clue about which skills they are looking for in a Data Scientist or why they even need one.
Most companies claim to be data driven but very few are. If you decide to interview with them, ask them for specific examples of how they used data to make decisions. If they cannot give you any, you will know it’s not true. So, unless constantly having to explain to people the value of your work is something you enjoy doing, you might want to pass!
I often see job descriptions that talk about insight. However, often there is not a clear goal behind the request but a vague ‘nice to know’ and, if it’s not scoped properly, the risk of being stuck in a never ending loop of ‘it would be interesting to look at X’ is very high. It is also very hard to quantify the impact of insight. It will exist in some director’s head and may or may not influence their future decisions. If this is all you’re being asked to do, chances are you are going to be a data analyst rather than a Data Scientist.
ONE. Data: No or poor data, no data engineering or infrastructure
You are applying for Data (!) Science positions. So Data will be your daily bread and butter and you want to ensure that it is available, accessible and organised. No data or poorly structured data can happen if you are joining a young startup that does not have a critical mass of customers yet or if it is a B2B company that relies on external partners for data. Without a basic infrastructure in place, a data scientist will likely be frustrated on a daily basis. Instead of contributing value via data science, you would be mostly dealing with data acquisition, organisation and pipeline building (which is the job of Data Engineers).
In order to not face any surprises before starting the job, here are some questions you should ask during your interview:
- What data is being generated or collected?
- How big is the current database?
- What is the approximate amount of data generated/collected each month?
- Where is this data being stored and how would I access it?
TWO. You are the only Data Scientist they are planning to hire in the near future/no head of Data Science
Hiring only one data scientist with no industry experience would be a big red flag for me. It is highly likely that the company is low on funding and can only afford to hire a junior Data Scientist despite realistically requiring someone with plenty of experience. Another scenario being that they hire several junior data scientists, but no Head of Data Science/AI, hoping that quantity will trump quality. This was the case at the company mentioned above where I ended up working for 2 weeks. My line manager had a psychology degree and was in charge of an NLP Data Scientist…
Questions to ask:
- How many other technical people will be on my team?
- Who would be my line manager and what is their background?
THREE. Role
The term “data scientist” has become a synonym for “data analyst” in certain companies. Some companies have deliberately rebranded their data analyst jobs to data scientists in the past, in order to avoid losing candidates to competitors. Another common scenario is data scientists having to be involved in data acquisition and organisation which is a data engineering job. So make sure you know exactly what will be asked from you and that your expectations are met. Don’t just reply on the job spec (as it is often very vague and general), ask the following questions:
- What should someone in this role deliver in the first 100 days? And the first year?
- Of the most effective people in this role, what key skills and behaviours do they have?
- How is success in this role defined and measured?
- What does someone have to demonstrate to progress to the next level?
FOUR. Roadmap: No/poor plan on how the team will deliver value to customers and the business
You are being hired as a data scientist so you would expect the company to be interested in gaining something from their data and generate value for their customers. So beware of directors who, if asked for their future plans, only talk about team growth, diversity, and upgrading the tech stack.
Questions to ask about the roadmap:
- How does the team deliver value to customers and the business?
- What is your roadmap and planned deliverables for the quarter? And the year?
- How does success look like for the team? What are the team’s KPIs?
FIVE. Tooling
Naïve as I was, I thought that every company under the sun where code is written uses version control. I was wrong and only found out on the job that they are only “planning to migrate to Gitlab”. Now I know that you cannot assume that version control is used, and now do you. And this is just one example. It takes considerable time and effort to learn new tools and to migrate all the code/data and people (!) to them. Thus, it is natural to have concerns about not using version control and/or using outdated, proprietary or inefficient tooling. Using good tools will make you more efficient and the work more enjoyable. So make sure to ask:
- What version control do you use (to give them the benefit of the doubt)
- What key tools does the team use in their day-to-day?
- What is your tech stack?
SIX. The job interview is too easy (e.g. no or very simple technical assessment)
If the interview does not include a part where your technical knowledge is assessed you are either interviewing for a role below your level or they do not have enough experience themselves. You want your superiors to be more experienced than you in order to learn from them and grow as a data scientist. Coding is not the most important skill for a data scientist, but it is something you will have to do every day. If this and your knowledge of data science is not assessed, you may not have any support on the job or that they do not program at all and use BI tools like Tableau and Excel. Alternatively, they might trust your CV so much that they do not need to test you which is a sign that they are desperate to hire.
The points above reflect my experience during interviews, however, here are a few more useful articles:
https://elu.nl/red-flags-in-data-science-interviews/
https://towardsdatascience.com/how-to-spot-red-flags-in-a-data-science-job-opportunity-22191e71f21
https://www.iqpartners.com/blog/is-this-employer-the-real-deal-5-red-flags-to-watch-out-for/
Good luck during your job search!
