Originally appeared in KDnuggets
Whether a data scientist is just beginning her professional career, or she is already a seasoned professional, working at a startup offers a number of advantages. Most startups are more hands-on, and usually most employees are involved in many aspects of the company. This offers a great opportunity for involvement and the expansion of skill sets.
At a larger company, one team might be responsible for Research and Development while another team is responsible for Quality Assurance. At a startup, the same team might be responsible for all aspects of research, development, and testing. As such, startup professionals are exposed to a wider array of experiences because they see the big picture and quickly gain the tools needed to become team leaders.
If a data scientist gets the opportunity to work at a startup, here are six tips that will help her and other data professionals succeed:
Six Tips for For Data Scientists to Succeed at a Startup:
1. Study the Notebooks
As a team player, a data scientist should review notebooks not only when called for a Code Review, but she should also read all of her team’s notebooks and side notebooks (including ones who might have a differing focus). This ensures she gains the fullest understanding of the entire process and how research is evolving. Data scientists make a myriad of decisions every day that they’re not even aware of. Studying notebooks allows her to reap the benefits of code review done by others at the startup.
The benefits are twofold:
- Tips, tricks, and hacks are nuggets of gold a data scientist can mine when studying the code review of her co-workers, thereby increasing her own productivity and enhancing her skill sets.
- Notebooks can reveal information that didn’t come up during standard code review or in meetings, challenging previous assumptions and shedding light on what next steps might be.
2. Check in with Teammates
By actively checking in with the team, a data scientist will not only have her finger on the pulse of what others are working on, she will find out how she can contribute in a meaningful and valuable way.
- Be a sounding board for others. Embrace becoming a rubber duck.
- Inquire about the challenges other team members are facing and offer to help if you are able.
- Get familiar with new tools, approaches, and problem-solving methods.
3. Embrace The DIY Spirit
Embracing the “do-it-yourself” ethos of a startup gives a data scientist the opportunity to develop software tools that in a larger company would be developed by separate engineering teams. This not only grows one’s skill set, it also makes one more self-sufficient, enabling one to troubleshoot and fix problems independently.
4. Take a Holistic Approach
At a smaller startup, each member of the team is a valuable part of a larger ecosystem working in harmony to create a product or service. This translates into the opportunity to be involved in and understand the reasoning behind every aspect of a company’s mission.
5. UI Through a User’s Eyes
Understanding the user’s interface is just as important as understanding the flow and the system’s stakeholders. When addressing a hypothesis, build an internal tool, train a model, a data scientist must consider:
- Who is using this model’s output?
- What does her/his usage look like?
- What is the impact of this intelligence task?
A Data Scientist can better understand what works with UI by seeking out user feedback, if it is available at the startup she is working at. Sometimes users might already know what works and what doesn’t so by listening to them it can help her decide what features to focus on developing. It’s also important to get feedback from teammates about possible blind spots, as their perspective will be different, and their vantage points will uncover oversights.
By taking full responsibility for what a data scientist creates, she can gain valuable insights offered by users and others on the team to good use.
6. A Pull Request is Just the Beginning
At a larger company, after testing a Pull Request, a data scientist job ends. At a startup, the same data scientist would go one step further and examine how the feature/model is performing live and how her insights are being implemented. Data scientists at startups must also pay attention to the live monitors and the logs, checking a few of them to spot any irregularities in the first few minutes/hours/days.
As these six tips demonstrate, success at a startup requires a greater level of adaptability, a willingness to improvise, and ability to pivot when needed. It also requires the desire to be a part of a tight knit team and see a product or service through every stage of development. Though the risk might be high, the reward can be great.