Writing a cover letter for a data scientist position

This applies both to sending out application as well as contacting people in the industry. A similar elevator pitch might also be useful during the obligatory 'tell me about yourself' question during an interview.

Whenever you meet people who already work in the industry, you have to be prepared to answer the four most common questions:

• Why data science?
• Why us in particular?
• Why should we hire you? 
• What have you done before?

It is surprising how many people stumble at this stage and it is important to get this right as this could be your first impression to a potential future employer.

Creating your back story

You will need to create a compelling back story, telling them how you ended up choosing data science as a career. Whether it is at a networking event or over the phone, no one is expecting a 30 minute monologue about your childhood. All people want to hear is that you are serious about this line of work and that it is nothing you just came up with browsing a list of the 10 most hyped jobs in a magazine.

Keep these tips in mind when you're creating your back story:

• Make your back story concise and don’t ramble - you should prepare a back story that is roughly 2 minutes long for formal situations, and about 20 seconds long for more informal situations
• Make it personal but not informal - talk about any links you have into the industry or the characteristics you possess that you think would suit insurance
• Show them you’re passionate - you don't have to know everything; passion goes a long way.

To keep your story concise, follow this simple structure:

• The introduction
• The insurance spark
• Your growing interest 
• What you want to do 
• Where you want to end up

The introduction

Tell them a little bit about yourself and then which university you went to or where you started your career if you are more experienced.

The analytical spark

Most kids want to be an astronaut or an actor when they’re older, rather than an office worker.

So tell them about what sparked your interested. Be honest – if one of your friends seem to love their job in analytics, tell them. Perhaps you had a statistics class in collage and were fascinated by the methods. Maybe are a programmer that needed to analyze data in some way.

Be as specific as possible and ensure your spark is personal. Being personal creates a connection and makes you more memorable.

Your growing interest

This is your chance to show them what you have done to satisfy your interest.

Talk about the modules at university you studied as a result, or the internships you’ve completed. If you are more experienced, talk about how your past work was related to analytics and how that fueled your interest. If you are a hobbyist talk about how you started reading on the internet and downloaded programs to make your own analysis.

Mention only two points at this stage, otherwise the back story will become too lengthy.

For each of the points, say what you enjoyed and tell them about something you would like to do more of or find out more about.

What you want to do

Now you need to state what you really want to do and what brought you to them. It never hurts to play to their egos, but don’t go over the top.

Your reasons should combine your interests, your background and what you have to offer them.

A common mistake at this point is to focus in to narrowly on the technical analysis part. Talking about the whole process from data collection to business value will set you apart. Use this opportunity to tell them what you can bring to the company, whether it’s your drive, passion or Qlikview wizardry, they want to hear what you can offer them, rather than the other way around.

Move on and tell them what interests you about the company. Again, be specific. Do your research and find out details about what makes their company different and what gives them their competitive advantage (everything they are proud about will be on their website).

Where you want to be

At this point be realistic. There are only a few people who make it to the heady position of CEO and there are only certain types of people who are really cut out for it.

It is best to stick to the medium term and state you’d like to be a develop as a data scientist, and learn how to work on real data in a business setting.

Show loyalty to the industry and an interest in developing personally. That’s all there is to it.
It is best to write it down, read it out aloud and edit it until it all makes sense and takes no longer than 2 minutes.

Example: Data Scientist for an E-commerce company

“I grew up in France but my family moved to Belgium when I was twelve. I was accepted into Brussels University to study Economics and spent four years there. During my third year, I went abroad studying in China.

During my forth year I came in contact with R as an analysis software when writing my master thesis and got interested in how I could program my own analysis scripts.

After reading on the internet about 'data science' I started the Coursera mashine learning cource and become really interested to make this my career. I am not analysis expert but by now I know how to take something from an SQL database, munge the data to the right format and do regressions or charts in R.

I had previously also managed a website so I started to dig in to the data from it making graphs of what visitors were doing from the server log files. I really like the process of figuring out the answer to a question I have from data and I also enjoy the technical challenge of programming. This is something I really like to work with full time.

I know you are seeing business intelligence as an important part of you organization and I would want to learn from skilled people so I in the long term can become a better data scientist."

How to get a job as data scientist having majored in Stats / Math / CS / Physics

Congratulations, you are the perfect candidate!

Probably you can't wait to go out there and do something. You have so many things you want to try, so many ideas for machine learning, modeling and AI.

Wait a minute, just to prepare you for reality. I say this once and for all.

70 % of any data science job, whatever it is called, will consist of cleaning data and getting it out of the database. I am serious. Pretty much anyone I have met during my career whatever type of business they came from, be they high frequency trading researchers or just the normal boring business analysts, spend at least 70 % of their day just getting the data in a workable condition.

Then when you have done this most answers will be obvious and with little need for any more advanced modeling than watching a chart or two.

Even where you will make something advanced, say a realtime classifier, you will probably just go with the simplest model. The reason for that is that in in general a more advanced model do not improve results enough, the better way to improve it most often to get better data.

So back to data cleaning you go again....

How to get a job as a data scientist with no relevant experience or education at all

This is obviously the hardest situation to be in but fear not, it is far from impossible.

Your first step is obviously to get your foot in the door, you can't be picky with salary and employers  but as soon you start to work your value will increase fast. Simply by doing something 8 hours a day you will soon become good at it.

The problem is therefore to get your foot in the door and that will be a little bit harder with no education or experience.

There are basically two ways to increase you marketability, getting some kind of education or learning things on your own.

I would only think about getting education if you fail at the steps below and in that case I would go for a shorter program.

Step 1. Decide on a reporting tool to learn

Your first job without any experience or education will be making reports for your boss or other people in the business. This is in general the first job for anyone starting out and a large part of any data scientist work unless you work only on a very specific problem.

It is also the one thing any data science organisation always want help in. In general the needs from management and other stakeholders are always bigger than what can be produced and data scientists in my experience would rather follow their own ideas than managements so any new hire that can take care of some of the dreaded management reporting is very welcome.

At the moment there are two tools I would care about, Tableu and Qlikview. That is because I think they are the coolest ones and any fun organization you want to work with have probably picked one of thoose two. I would not care to learn things like Business Objects unless you like wearing a suit and want to work as a consultant. I would neither care to learn stuff like R and SAS since these are best at making statistical analysis.

Step 2. Download it and learn it inside out

You are lucky that both theese two software packages has free versions. Download them and learn them inside out. Youtube is your friend.

Step 3. Learn some SQL

You do not need to be a master. Just enough to join some tables and do a GROUP BY.

Step 4. Make a good looking dashboard that connects to an SQL database

Here get as creative as you can. You want to make something you can show prospective employers. It could,

• Show a timeseries in some fancy way, preferably with a compared to last year or last month line.
• Have a table that shows both aggregates and daily and monthly results.
• Some kind of interactivity like the possibility to break down into subgroups
• and whatever you can come up with, the above are suggestions since it is something that comes up in almost all business reporting.

Come up with data if you can't find anything.

Step 5. Learn some Excel basics

I am talking vlookup() and pivot tables. Again youtube is your friend.

Step 6. Get this in front of employers

By now if you have showed this to an employer you will suddenly seem like a better value than the newly minted collage graduate that hasn't done any of these things before. While probably not an analytical superstar they will see that you can bring value from day one and that you are motivated, specifically, here is someone who can do this boring reporting for me while I do the stuff I find fun.

From there you can hopefully advance.