What does it mean to be a Data Science PM?
About us: Andrew (AM) and Chandrika (CM) met during their MBA program at MIT Sloan and connected over their passion for product, growth and paying forward the help they got transitioning into product management. AM currently works as a Product Manager at Moveworks and CM works as a Product Manager at DocuSign.
In this new series, we are covering our peers in PM roles across different organizations to give you a view into what various roles demand and which role is likely the best fit for you. For this post, we spoke to our friend Jasmine who is currently a PM at VMware. Before her MBA, Jasmine used to work as a Data Strategist.
Jasmine currently works at VMware as a Data Science PM for VMware Cloud. Her work revolves around managing cloud capacity needs for all customers as VMware Cloud scales.
In this post, Jasmine reflects on her experience working as a Data Science PM.
How do you define the business objectives that you are working towards with your project?
Capacity in the cloud is one of those things that you never want your customers to have to think about; so our main business objectives revolve around maintaining a seamless customer experience (not have customers worry about capacity) while ensuring that we can support more customers’ capacity needs in a way that is cost-effective for the business.
Drilling it down further, the two main objectives I focus on are balancing minimizing costs while minimizing instances that customers experience capacity constraints; these objectives are often in opposition, but are also dependent on each other: minimizing costs too much may adversely affect our business’s ability to support our customers at any point in time, and too much capacity is very expensive to the business.
To measure our costs, we track unused hardware over time, and to measure the impact on customer experience we track instances of capacity constraints that affect customers. To achieve both of these business objectives my team focuses on building improved machine learning models to more accurately forecast the capacity needs of the business in the future. As part of this effort, we are naturally focused on mitigating the biggest risks in our forecast, which are usually host deployments in smaller data centers and/or large magnitudes of host increases in a short time-frame. We measure the dependence on customer input in forecasting to understand how much risk is in our forecast at a given time.
Over the last year, the pandemic has brought a lot of growth for VMware Cloud; in March, there was a huge demand to move to the cloud as companies were forced to move online in a matter of days. Ensuring there was enough capacity for customers was one of the biggest priorities of the business. The work my team focuses on as the pandemic persists will continue to help VMware Cloud balance these two business objectives with intelligent forecasting, processes, and management levers as it scales. Data science is at the center of the work that we do.
How do you go about defining requirements for your projects?
A large part of my process in defining requirements is to first deeply understand the problem I’m trying to solve. Usually this involves talking to different stakeholders (Finance, Customer Success, etc.) to understand their pain points, why current processes exist, and identify if/where there are bottlenecks in processes. From there, I usually create a product requirements document and then send it to review by my engineering and data science team. I find that writing out requirements fully and getting feedback early is critical to making sure that my entire team understands the problem we are trying to solve; this usually prevents running into unexpected issues down the line. Lastly, I make sure that I outline the impact on the metrics of the business objectives I mentioned earlier in order to clearly define the impact of product improvements and get buy-in from different teams.
In data science, requirements are often subject to change because of the frequency of feedback cycles. Our team iterates at a pretty fast pace. That being said, we structure our quarterly planning so that work is time-bound and there is more flexibility for change and iteration. I’ve also learned that writing comprehensive and valuable data science requirements often requires the PM to do exploratory data analysis and have conversations with data scientists first. This pre-work is key to making progress.
How do you prioritize projects?
Our VP and Directors share with us the strategic priorities for the fiscal year, and then each PM sets their own priorities, using these big-ticket themes as a guide as we create our quarterly plan. For my own quarterly planning, I also use my own simple framework of impact/confidence/cost. With many different areas that require data science work, prioritizing is key to using the time of my data scientists wisely; having clear requirements and consensus on a topic are critical and are a major input in the confidence portion of my prioritization framework.I have found that utilizing the grooming sessions in our SCRUM process wisely leads to better requirements overall. As usual, priorities of the team are constantly being assessed so it’s important to have flexibility in planning and a good relationship with my engineering manager.
Overall, who do you think will enjoy the role of a ‘Data Science PM’?
A background in data science is definitely helpful, but not necessary. A strong desire to explore data, run experiments, and distill insights is key to being a successful Data/Data Science PM...and enjoying the job! If you like storytelling with data, this could definitely be a PM role to consider. Also, there are constantly new tools/services/frameworks that are available in the data science world, from machine learning techniques to new visualization tools. It's really important to be a constant learner and stay up-to-date. I would say that others skills necessary to be a successful Data Science PM are very similar to those required of other PM roles; context switching between strategic and technical problems, collaborating and building relationships with other teams, and consistently prioritizing what is the best for the customer with the resources you have are all key parts of the job.
If you found this valuable and want to support us, consider buying us a coffee!