More than a year ago, I joined Siemens as a data scientist.
Of course, I was happy when I got the offer, but also somewhat hesitant: I knew that the team was just being ramped up and I would be the only data scientist. This would mean both opportunities and challenges.
Besides a few internships, It was my first full-time job after graduating with a Master’s degree in computer science. However, I had done multiple machine learning projects as well as software engineering internships and I felt up to the task.
Once I joined, I was eager to meet some other data scientists within my business unit (which had around 6000 employees, after all). I asked around and did some research. After a few days, it was clear. I was the only data scientist in the whole business unit.
The Good: Responsibility and Opportunities
In some sense, it was a great opportunity for a first job. Being in charge of data analytics and machine learning in my team, I got to do all the steps of the data science process myself: Exploratory data analysis, visualizing the data, talking to domain experts, testing hypothesis, modelling and writing the production code for our solutions. Responsibility comes with visibility, both internal and external. Only a couple of months into the job, I already pitched the analytics component of a project to customer management. On other occasions, colleagues I did not even know contacted me and asked for advice on data analytics. After all, I was the machine learning expert in the building.
For these reasons, being the first data scientist in a company is a great opportunity to network. Getting to know colleagues from different parts of the organization not only helps you advance within the company, if you want to stay there. It also helps you find new jobs
Responsibility certainly is challenging. At times, you might feel out you depth, like you are an impostor that just tricked this unsuspecting team in hiring you. Recognize these situations as what they truly are: Opportunities to grow. Now you can deepen your knowledge in different data science skills in a real-life setting: From modelling to implementation to communication you will be faced with a wide array of problems that you have to solve. You can do it, even if you have to got the extra mile and read up on some algorithm or software design pattern. Use this opportunity to gain knowledge about different aspects of data science. For instance, I studied time series modelling and forecasting. My focus has been computer vision previously, but now I mostly had to deal with sensor data in the form of time series.
For many data scientists this could be an opportunity to improve their software engineering skills. I met some data scientists, who are experts in computer vision or natural language processing or have an impressive knowledge of statistics, but at the same time are not too familiar with systems design or deployment of machine learning models. Depending on their previous experience, they might also not know the best practices of software development. Now is the time to read up on version control, testing, code review. Ideally, you are working with an experienced software engineer. Don’t be afraid to ask them or talk with them over lunch about these things. Software engineers love to talks about programming languages, software architecture and branch policies for hours on end.
Besides technical expertise, it is crucial to make an effort to communicate effectively. Communication is always important, but if you are the first data scientist in a company, it is super important. Many people will not be used to your way of thinking about problems or might have to be sold on trying a data-driven approach. Others might have read some article on machine learning and AI and have inflated expectations that you have to manage. Lay out what is possible, what may be possible, and what is not (yet) possible. Communicate your ideas and results with as little jargon as possible. Explain the algorithms and techniques that you used and why you used them such that a layperson will understand.
The Bad: No Mentors or Peers to Learn from
As I said, the responsibility that comes with being the only data scientist in your company forces you to be a generalist and provides a lot of visibility. It is an opportunity to learn — to delve deeper in some topics and brush up on others. However, I think I could have learned some things faster, if I had more experienced colleagues to guide me.
As the only data scientist I did not have day-to-day contact with other data scientists to learn from or to bounce ideas off. At least not in person and without scheduling a phone call beforehand. I have been mostly involved in with a single project that started shortly after I joined Siemens and is currently approaching a major release. Through most of that time, I have been working remotely with two data scientists from an IT consultancy that we collaborated with for this project. I get along well with them. The cooperation has been very open and we have never treated each other differently than you would treat a colleague from the same company. Each of us has different backgrounds, strengths and interests: I, for instance, am a computer scientist by training and a programmer at heart. After the initial phase of exploratory analysis, I naturally gravitated towards implementing our models, tools and pipelines.
However, they do not sit in the same office, not even the same country. In fact, the team working on the project has been highly distributed from day one — and belonged to two different companies. Due to bad IT infrastructure, communicating outside our daily meant a lot of overhead for a long time: Emailing them, scheduling a meeting, joining the call. So, still, no peers or mentors.
The Unexpected: Being an Interviewer (It Is Hard)
A few months into the job, my manager decided to hire two more data scientists. I was excited. Not only would I finally have some peers to work with and learn from each other, I would also participate in the interviewing process. After only a few months in the job, fresh out of college, this felt like an honor.
Our interviewing process was quite straightforward: Candidates had an interview with my manager and myself. I focused on the technical aspects, data science, machine learning and programming, while my manager asked more general questions about motivation, mindset or leadership qualities. If we decided to move forward, we would invite the candidate to give a presentation about a project they had worked on.
Interviewing is hard — everybody knows that. I learnt soon that is also applies for the interviewer. Assessing whether someone is the right fit for a position in a short period of time and after just a few questions on machine learning, algorithms and programming is really hard.
For the next interviewing cycle I (or my new colleagues) will definitely spend more time with the candidates. You have to figure out a lot before you can be confident that you found a good hire: Knowledge and skills, ability to learn and solve problems, cultural fit. And you really want to avoid false positives — after all, they will be your coworker for the foreseeable future.
The Takeaways
There are a lot of companies in traditional industries that know seek to ramp up their data science capabilities. Therefore, like me, you might be the first and only data scientist in your organization. This is an opportunity! If you communicate effectively, you can grow your network quickly and take on a lot of responsibilities. Certainly, this is an opportunity to learn different aspects of data science.
If you don’t want to jump in at the deep end, ask upfront if there will be other data scientists within your team or your organization. And if you find yourself in a situation that you joined a company as the first data scientist and you are feeling lost, ask your manager about the plans to build a data analytics team and how you can help. They will certainly appreciate it. And if there aren’t any plans or you want to leave for a different company with more data science expertise, come up with an exit strategy and in the meantime use your time wisely to grow, learn and develop a network.