In the 1970s, NASA developed the Technical Readiness Level scale to measure the progress of nascent technology. Since then, many other engineering bodies have developed similar scales. In most TRL models, readiness levels increase from basic research (lowest numbers) to commercial availability (highest numbers). This is useful not just for tracking progress of technology development, but also for setting realistic expectations for work output.

In web-centric engineering spaces, TRL models are not commonly used. The reasons for this are manifold, but the reality is that most projects start well above the basic research levels. However, as we seek to integrate Data Science into our web-space projects and apps, we find ourselves doing more fundamental research to understand just what the data tell us and how they can be used. This can be difficult to reconcile with continuous improvement/continuous delivery project management models. The TRL scale may offer some insight on how to set expectations and track progress.

By thinking of Data Science as a "full stack" problem, we span the TRL scale from basic research (e.g. exploring our data to see what patterns may emerge, developing new algorithms) to commercial release (e.g. shipped features). Instead of treating Data Science as a standalone team that interacts with product teams at-a-distance, we can use TRL scales to intelligently integrate data science with other teams, from front-end and back-end engineering to QA to UX to customer relations. The TRL model also allows us to identify strengths and weaknesses among scientists so we can focus on how to allocate time, effort, and training to put together more effective data science teams.

This talk will break down a basic TRL model, explore what data science might look like at every step, and how to use it to fill out a true "full-stack" data science team.