What Kind of Projects I do as a Data Scientist
--
I often get asked about what I actually do as a Data Scientist and how different it is from working as an Analyst or a Data Engineer. First of all, every company’s JD of a Data Scientist is unique. Some DS jobs require more technical knowledge, some require more analytical skill and so on.
I started in this role as a Senior Analyst for only a couple of months and then later got moved to being a Data Scientist. I guess some companies, including mine are still trying to figure out what the differences are between their teams.
So to explain it better, here are some of the high impact and regular projects that I’m currently working on— without of course, giving away too much information that I’m not legally allowed to disclose.
High Impact: Predicting Product demand to achieve $$mil+ weekly inventory cost reduction
One of the projects I work under this category involves a Machine Learning model for time series forecasting that predicts demand for certain products. On a day to day, this project is heavy on monitoring daily predictions, data engineering pipelines and monitoring actuals. In some cases where weekly targets are not met, root cause analysis may be required and this would involve doing some statistical analysis or forecast analysis. The impact of this machine learning model is on majority (over 9k+) of the company’s stores in the US.
High Impact: Creating Algorithms to improve efficiency in a Supply Chain Network by $$mil+ annually
An example of this project can be, doing research about the end to end process of a specific supply chain network, looking at different inventory, store and a few other metrics. Identifying the business problem, coming up with targets, then eventually creating mathematical formulas and an algorithm (written in python / R) that will support in achieving the business goal. In most cases, algorithms created are hardcoded from scratch so most of my time is spent on analyzing scenarios that can potentially break an algorithm, debug codes and test multiple scenarios. Some mathematical linear equations and statistics are involved.
