So, it finally happened. The dreaded day I feared, but was not entirely a surprise. Due to financial restructuring I was let go from my position at 25:2 Solutions. It's been a wild 5 years and I am grateful for the experience that I had with the company. After all of that, I figured I would share some lessons that I have learned from this unique position.

Learn to Communicate Well

A massive barrier to productivity is poor communication. When I first started working with 25:2, people would ask for specific data science tools or use terminology that was known. Initially I made the assumption that they, being the Subject Matter Experts, were well aware of their needs. I would spend time creating a tool or software to solve the problem as described and on presenting the solution, the response was less than expected. After a few times, I learned to ask clarifying questions like: "What goal are you trying to solve with this tool?" And "how do you imagine the end-user interacting with this?" Clarifying questions are important. Another important piece of advice is coming to a joint understanding of the problem and framing it as a statement. Both you, your team, and the stakeholders should agree on this. Consider framing the problem as a SMART goal, or at least where there can be a roadmap of SMART goals that lead to the solution.

Optimize Judiciously

I had a professor who taught programming that would say this. Without trying, programmers can enter a destructive cycle of perfection and completely forget about the end goal. The successful tools of today weren't perfect the first time they were released. Aim to solve the problem, then perfect the solution.

Say Less

This is a serious problem with scientists. We err on the side of verbosity as we want everyone to know how much effort went into this project. This approach leads to the dilution of the message. You might have worked really hard on this and tried 13 different analysis methods, but the final result the important part. One person I worked with years ago went into his presentation and immediately the business people left the room. He would complain that no one took his presentations seriously but when you're in a room full of business people, say less. Become an experienced storyteller and that will serve you well.

Aim for Simplicity

Often as a data scientist, you may be drawn to the most complex analysis method. But complexity is not a guarantee of success and choosing more complex methods without analyzing the distribution, covariance, or other simple tests can be a complete time waste.

The Burden of Support

When creating new tools, it can be considered that at the creation of that tool you're done and can let it ride. As if creation was the only burden. This is quickly realized to be incorrect as the support requests start pouring in. Bug reports, feature requests, security patches.. The need to support software tools is ignored at your peril. If you create a tool, consider the support burden you will have as more people start using the tool. This is especially true at a small company where you wear many hats.

Closing

Finally to my coworkers who I had the pleasure of working beside, I will miss you all. I hope that I made your time there better and lightened your load, as I know you all made my time brighter. I enjoyed the experience and now am in a unique position of looking for the next opportunity to make a difference.