As Data Scientists at Jana, it’s our job to explore ways that data can guide our company’s overall strategy, as well as create better experiences for our users. Through designing and implementing models that mine our massive datasets, we learn more about our users and improve their experience. These improvements include serving relevant and targeted recommendations, to detecting and blocking fraudulent users.
Over the last several years, I’ve had the pleasure of consulting and working for companies large and small across different industries. I’ve learned that startups are a particularly fun and rewarding workplace for data scientists for three big reasons:
Initial analysis can have substantial impact
You can ask your own questions
1. There is an abundance of low-hanging fruit
The proven value of data has made data collection a top priority for most startups, across all industries and stages. However, many startups make data collection a goal without really knowing how to eventually use that data. As a result, at a young startup, you’ll likely find unique and rich datasets that nobody has looked at before.
Since their release, our products at Jana have been generating large streams of rich data. This includes behavioral data on our users—such as their browsing or search history—network data, and inferred data, such as user interests. Even with roughly 10 full-time data scientists, analysts, and engineers, we are still scratching the surface of our datasets.
2. The initial analysis will have substantial impact
A data scientist can create a working prototype quickly. Iterations to improve that prototype, however, can take much longer and will typically only marginally improve performance. As a result, a data scientist at a startup can quickly create new high-impact prototypes. Whereas a data scientist at a more mature organization will likely have to spend long periods of time maintaining or marginally improving the performance of already existing systems.
Before joining Jana, I worked with a fashion industry startup on how they could help consumers find out how a piece of clothing would fit them before they made a purchase. I used their unique dataset to create an algorithm for fitting clothing to body measurements. Within one day of work, the algorithm outperformed human experts. But, if I went back to improve upon that algorithm and add to the analysis, the work would be much harder and complicated than it was the first time around with fresh data.
3. You get to ask your own questions and solve your own problems
At small organizations, you’re more likely to be the only, or one of the few, data scientists. As a result, you’ll often get to decide what’s the most important problem to solve or what questions you want to ask.
Data science, when done correctly, will involve long periods of confusion and uncertainty. It requires a certain skillset and mindset—the ability to be make progress despite the confusion. Being able to decide what questions to answer and what to be confused about will certainly make one a happier Data Scientist.
If you’re are interested in joining a data-driven startup in Boston, Jana is hiring.