Why we launched this scheme

Due to Peak’s continued growth, our data science team has been almost doubling in size each year – we started with one data scientist five years ago and now have almost 70! We hire a lot of great people straight out of university having just finished a BSc or MSc, so this year we decided to offer a more structured training programme to give new joiners the best possible start to their careers in data science!

The recruitment process

We received 210 applications for eight roles, which was an amazing response for our inaugural scheme. We advertised the graduate scheme on LinkedIn, the Peak website and also got in touch with some of the university contacts we had within the team which really helped to drive applications.

The initial application process was for candidates to upload a CV and to answer four questions telling us a bit about themselves, their interests, and why they want to work as a data scientist at Peak. Asking these questions really helped us to evaluate the applications better than if we just had a pile CVs to trawl through – particularly as most applicants had strong technical undergraduate degrees, making it very difficult to decide who to choose!

We narrowed the 210 candidates down to 28 who were invited to take part in the next stage – allowing for several drop outs as we expected graduates to be interviewing at multiple places at the same time.

The next stage was a technical challenge and an assessment day. The technical challenge involved analyzing and building a predictive model with a chosen data set. We decided to limit the challenge to one day (sending it at 09:00 with an 18:00 deadline) to prevent people from spending too long on it. We also ran it on a different day to the assessment day to give the candidates enough time to properly understand the assignment and give it their best shot!

We held a virtual assessment day that 15 people from Peak helped to run. We had a variety of tasks throughout the day to challenge the candidates in multiple areas. These included some quick-fire questions (e.g. What’s your superpower? What’s your main strength?), a deep dive into their approach to the technical challenge, a group presentation and an individual interview. It was an exciting day that not only helped us to select our frontrunners, but also meant that the graduates had a chance to meet and chat to lots of data scientists from Peak!

The graduate scheme

We’re keen for the graduate scheme to give an all-round view of what being a data scientist is like, so that they can develop a range of skills and also plan what career path might appeal to them in the future.

The graduate scheme is 12 months in total and includes:

• Four three-month placements across our retail, manufacturing, insight and R&D data science teams. Here the graduates will learn what data science is like in practice, from working with our customers to developing tools other data scientists can use
• A group project that they’ll work on together from beginning to end, so they can see how our projects usually unfold
• Graduate training sessions every two weeks that will build a solid foundation across both commercial and technical skills, as well as the opportunity to attend other data science training sessions!

The graduates

This week we’ve welcomed eight graduates into the Peak Data Science team. They have studied a variety of degrees including Business Analytics, Medical Sciences, Operational Research, Maths, Physics and Machine Learning. Between them, they also speak eight different languages!

We hope they have a fantastic year at Peak and we can’t wait to hear about everything they’ve learnt by this time next year!

From our experience, there are three key areas of skills you will need to develop in order to get a job in data science; commercial, theoretical, and programming.

Currently you may have limited experience in these areas, but this blog post should provide some resources and ideas to help you practice and get you one step closer to landing that first data science job!

Commercial

In business, the role of a data scientist means that you get to work with lots of different people; from engineers to sales teams, as well as your customers. This means that you need to develop your commercial skills in order to communicate the impact of data science and be able to explain your work to others.

Practice presenting in front of others. Find opportunities to communicate or present a technical project to a non-technical audience. This is something that we do day-to-day at Peak with our customers. It’s easy to slip into technical jargon without realizing, so the more you practice, the better!

Data visualization. In an industry role, charts and visualizations are a useful way to communicate findings in data, and it’s important that you can make high quality charts. ggplot2 is a popular charting library in R that you can practice using. Data visualization is a huge topic in general, so visiting sites like Information is Beautiful can be great inspiration

Turning a real world problem into a data science problem. Projects you may have worked on so far might be very well defined and use small, clean data sets. This often isn’t the case when working in the industry! Kaggle is a good place to start for finding many different datasets and tutorials. Once you’ve gotten comfortable working with data, try to solve a new problem that you’ve thought of yourself.

Theoretical

When applying data science models to commercial problems, it’s important that you understand the mathematics underlying these models. This helps you to understand if you are solving a problem in the best possible way, and will help you to improve your solution if your model isn’t performing as well as expected.

You need a good mathematical foundation – ideally some statistics. You don’t need to have studied maths; other similar subjects like physics and engineering are also great preparation for a job in the data science industry.

Being able to explain how models work is key. As an example, if you’re using a particular algorithm (e.g. Random Forests) or model performance metric, you should understand how they work and why you’re using them. Data scientists at Peak are encouraged to always be curious, and an understanding of how things work enables you to spot risks and address them quickly.

If you come from a less numerate background, it may be worth doing an MSc in data science before applying for jobs. Several MSc programmes have industry placements to give you some commercial experience while you study. We have hired several graduates from the Data Science MSc at Lancaster University.

Programming

As a data scientist, you need to be able to put your mathematical and commercial knowledge into practice by writing code. You don’t need to be a fully fledged software engineer, but it’s important that you are comfortable with programming and open to learning new languages and technologies!

Some experience with R or Python is expected in most data science jobs, so it’s good to learn at least one of those before you apply. Some useful resources are R for Data Science for R, and Codeacademy and Numsense for Python.

SQL can be useful, too. It’s not usually a prerequisite but often used a lot in data science jobs. W3Schools is a great resource for this.

If software engineering is something you’re interested in, data science teams often need people who can generalize code and create packages for the team to use. Chip Huyen’s blog is a great place to learn about machine learning engineering.

We hope this blog will help out any aspiring data scientists and offer some practical advice on what you can learn before applying to a job in data science. When you’re ready, make sure you apply to Peak! 

Amy and Sorcha are Data Science Team Leaders at Peak. Got a question for them about how to get a job in data science? You can connect with Amy on LinkedIn here, and find Sorcha here.