Like so many other organizations, schools across the US, from kindergarten through college, scrambled in March to move from what has almost always been an on-premises educational experience to something new -- some kind of online learning program.
Some schools went with recorded online tutorials and set up office hours for students to "meet" with teachers, via phone or video conference. Other schools conducted live video conferenced classrooms and discussions. In some cases, schools shut down their on-premise teaching for the rest of the school year immediately. In other cases what started as a 2-week closure turned into the remainder of the school year -- months of instruction.
But at least one school degree program was doing the whole online learning thing really well from the moment that first stay-at-home government shutdown order was issued. That's because the University of California Berkeley School of Information Master of Information and Data Science Degree program has been online only from the time it began in January 2014.
Unlike MOOCs (massive open online courses), the UC Berkeley Data Science program was created in the image of a typical advanced education course, just delivered remotely. The school has the same small classes and discussion groups you would expect from an on-campus school. The expectations of students are also high.
"Just because it is online doesn't mean that you can disengage and not pay attention to the discussion," said Kyle Hamilton, a lecturer with the program who teaches four sections of Machine Learning at Scale for the program each semester. Hamilton is also a graduate of the program.
Over the last 6 years since its inception, the program has gotten smarter with its technology, refining the tools it uses to connect with students and the tools students use to do their work. The program is remote and attracts students from all over the world, yet most students still hail from the San Francisco Bay area where companies like Google are based. UC Berkeley School of Information has often partnered with tech giants to provide tools for student use. For instance, Google would grant credits to students to use data science tools in the Google cloud.
But just because it was free doesn't mean it was easy. Students would need to use the credits to set up their own infrastructure to complete the coursework, yet the setup itself was often challenging. You could argue that dealing with the frustrations of setting up the technology is something the students would need to learn in order to deal with tech in the real world after graduation. But the frustrations took away from time students had to apply to the actual course content in Hamilton's course -- Machine Learning at Scale.
"When the course started, there were so many moving parts," she said. "Students were expected to do their own infrastructure set up. What I've tried to do over the semesters is to simplify that."
The program has used several different clouds over the years including Google and AWS, "but I cannot personally support all these clouds," Hamilton said.
In the most recent semester, however, UC Berkeley School of Information has been piloting a new partnership with cloud-based big data platform provider, Databricks, a company that got its start as a cloud-based provider of open source Apache Spark. It has since expanded its cloud-based platform to incorporate a full host of open source data science tools. This month Databricks made the college partnership program official, announcing the launch of the Databricks University Alliance, a global program offered to educational institutions at no cost to help their students graduate with the skills they need to land jobs in data science. In addition, students working from home on their own can log in and use the free Databricks Community Edition and access educational content from the company.
In conjunction with UC Berkeley School of Information, the alliance program also provides students with access to tutorials, content and training materials on open source tools including Apache Spark, Delta Lake, and MLflow. It is powered by public cloud providers such as Microsoft Azure and Amazon Web Services (AWS).
The tech support on the infrastructure was a huge help for Hamilton, who worked with Denny Lee and Rob Reed of Databricks to provide students with a stable environment to use.
"I've had good reviews from students about the infrastructure," Hamilton said. "They took a list of my students to onboard to the platform. They provided technical support. If I had questions, I could email Denny and Rob and they would get back to us within a day."
Such partnerships between educational institutions and tech companies go back a long way. Apple famously offered special packages and discounts to schools in the 1980s and 1990s, making the brand the choice of a generation even after graduation. Google is now leading in that space with so many schools choosing inexpensive Chromebooks as a way to deliver curriculum to students.
The UC Berkeley School of Information program and the Databricks partnership together target a subset of advanced education students. But such programs could be a forerunner of a new way to deliver education in a post-pandemic world. Maybe a new class of elite cyborg universities will emerge. Such programs could disrupt the education market by being less dependent on physical location and potentially enrolling many more students at a lower price after an era when many graduates spent decades trying to pay off massive college debt.
In any case, the students in Hamilton's Machine Learning at Scale class will have an easier time setting up their infrastructure so they can focus more on the material they are there to learn.