Migrating Learning Technology Solutions to the Cloud
The first time I stepped on the Boise State campus was as a first grader in 1982. I enrolled in ‘96 and began my IT career as a student employee a year later. Many things have changed since my first job when I swapped token ring cards for ethernet and upgraded many PC BIOSes in preparation for Y2K. For instance, we no longer refer to ourselves as “BSU” (confusion with Ball State University, and “Boise State” just has a more definitive and unique sound), and we are ramping up our IT staff and business units to prepare for revolutions of machine learning and artificial intelligence (AI).
I sense we are not alone in our quest to unlock the potential of machine learning and AI for our organization. Public cloud providers are abstracting user accessible interfaces from the complex, underlying web of big data, machine learning, and data warehouse technologies required for machine learning at a rapid pace, and the successful use cases arrive as fast as the incremental developments in machine learning (ML) and artificial intelligence (AI) technologies.
In many ways we are still at the beginning of our public cloud journey
Last year, I spent time during the holidays studying for and passing an exam endorsing myself as a cloud practitioner. I admit that the allure of proving to myself I could still pass an exam was as much of a motivator as learning the basics of the public cloud; the base knowledge propelled me to learn more about big data, machine learning, and the differences and dependencies between the two. New developments in the past year have led to new services along with a reduction in barriers to entry. Time series forecasting will soon be a matter of supplying the prebuilt algorithms with data. In Boise State’s case, data that will “hopefully” lead us to predicting instructional and administrative capacity necessary for upcoming semesters.
However, the journey that began with a growing understanding of the development of machine learning technologies would not produce results at the snap of a finger or an organizational edict. After all, this is higher education, where the roots of the academy manifest themselves in a highly distributed power structure. It’s not an impossible structure to navigate, but it does require patience, persistence, good relationships, and a bit of governance. These factors all lead to time, of which higher education institutions have a historical abundance; although growing expectations from students, parents, regulatory entities and others drive us to be more business-like, speed does actually matter.
Macro level organizational challenges are one thing, and creating a culture of adopting and implementing new and disruptive technologies creates micro level cultural challenges as well. Boise State has made progress using public cloud services in the past two years, but I will admit I could have done a better job communicating our strategy that the public cloud creates benefits and opportunities, while also tamping down the fear that we would use the public cloud for everything up to, and including, staff consolidation. Boise State has a history of successfully adopting and adapting new technologies (and yes, job duties changed) but we have not given walking papers to any IT staff due to these changes. For instance, our Learning Technology Services group will have completed a multi-year migration of all learning technologies to SaaS products by the end of 2018. The team has not experienced staff reductions; on the contrary, the team has grown and will continue to grow throughout 2019 as they absorb responsibility for campus testing services. The work of the individuals has changed, but the value of that work has increased, raising the profile of the team and our department.
While the transformation of our Learning Technology Solutions team is nearly complete, in many ways we are still at the beginning of our public cloud journey. How do we refine our approach to the public cloud in a way that produces meaningful outcomes and begins to change how we perceive the threats and promises of public cloud services? And, how do we communicate our intentions to the areas that can be partners in our progress as opposed to detractors to success? The term “change management” is often used, but this is a classic stakeholder management problem. Looking at the challenge through the lens of managing the expectations, concerns, and hopes of those impacted by a project (even a proof of concept project) is the basis of stakeholder management, which extends to the hierarchies and branches of an organization as well as technology teams. Our staff have a stake in the success and promise of new technologies, but sometimes pioneers are hard to find.
In the past year we have reinvested, focused, and clarified the operational goals of our Research Computing team. Through the process, I’ve learned about the method of finding researchers to apply for grants. The generally accepted term is “call for proposals.” Essentially, the call is a broad announcement from various organizations, including government agencies, foundations, and even corporations, to solicit solutions to problems within a sponsoring organization’s predefined requirements. The structure does not tell the researcher how to solve the problem, but it does provide the problem to be solved as well as parameters that must be adhered to during various stages of research. Typically, the call also provides review criteria and may also define expected outcomes--the factors for success. This model is so successful at attracting researchers to compete to solve problems that universities pride themselves on the level and intensity of research produced by their institutions. What if the call for proposals concept were revised to be a call for participation to those interested in using new technologies, such as data lakes and machine learning, to solve a problem for a business unit in need of making timely, accurate, and labor-free (or near free) decisions?
In September, we announced a Call for Participants: New Technologies in our Organization - Data Lakes and Machine Learning. Through advertising, promotion, and serendipitous conversations, seven individuals from our IT organization and three from our online learning program responded to the call. Some participants wanted to stretch their skills, which is why our Solutions Architect is honing his project management expertise and our Web Developer is learning about ML deep learning models. For others, the call is about extending their abilities; if one can use our integration tool to move data between on-premise applications, how is the process different when the data target is in the public cloud? We also have exceptional representation from our online program area that has defined the objective: use machine learning to automate existing data collection and processing tasks to improve the accuracy of summer course capacity planning.
The business representatives keenly understand their data and have an intermediate understanding of machine learning technologies, and since October, the team has come together and a articulated a business problem. On-site data lake and machine learning training is complete, and we have balanced the project workloads to respect a pre-project agreement with department heads stating the project would require about 5 hours of project time per week. We are optimistic the project will finish by the end of February. Then we will evaluate where we succeeded, where we need to improve, and whether subsequent calls for participation can further align our technology organization with the business units we support and the students we serve.