Labz 'N Da Wild 2.0: Teaching signal and data processing at scale using Jupyter notebooks in the cloud

Demba Ba discusses two new signal processing/statistical modeling courses he designed and implemented at Harvard, exploring his perspective as an educator and that of the students as well as the steps that led him to adopt the current cloudJHub architecture. Along the way, Demba outlines the potential of architectures such as cloudJHub to help to democratize data science education.

Demba Ba recently designed and implemented two new signal processing/statistical modeling courses at Harvard that leverage Jupyter notebooks in the cloud to bridge the gap between EE (data collection) and CS (data processing/management) education—a framework that he calls Labz ’N Da Wild. The backbone of Labz ’N Da Wild is a scalable, cost-effective implementation of JupyterHub on the cloudJHub, which addresses some privacy concerns associated with cloud-based technologies. CloudJHub launches Jupyter Notebook coding environments online for users and requires no installation—users simply log into the website and immediately have access to the Jupyter Notebook. Each user gets a dedicated EC2 instance, created when the user first logs in. A JupyterHub cluster manager automatically stops EC2 instances that have been deemed inactive. The cost of a user active on cloudJHub for 30 hours in a given month is $2. Demba discusses these courses, exploring his perspective as an educator and that of the students as well as the steps that led him to adopt the current cloudJHub architecture. Along the way, Demba outlines the potential of architectures such as cloudJHub to help to democratize data science education.