How our Data Science Department chooses what to research
One of the biggest challenges in a Data Science organization is identifying and prioritizing the most promising directions to undertake. Here’s how we create our backlog and place our bets.
At Riskified, we have an extremely capable data science and research department, composed of over 30 data scientists and analysts. As a fast-growing startup we take a practical approach, measuring ourselves based on the actual business value delivered every quarter and how effectively we interpret data (more about it in my previous article). We use the Agile methodology and our own adaptation of Scrum to keep things moving at a high pace. Additionally, much of our team has grown in the fraud domain and we have a keen understanding of the business. This means that the majority of our quarterly planned work is driven by internal research initiatives with the minority coming from business stakeholders in other departments.
One of the biggest challenges in any quarterly plan is identifying and deciding on the most promising research directions to undertake. While we can’t say that we’ve got it all figured out, this blog will try to shed some light on the tradeoffs we consider and how we tackle this problem.
