Business opportunity
Master of Business Administration (MBA) students at the University of California Los Angeles (UCLA) Anderson School of Management need to navigate an overwhelming amount of university resources designed to support the transition into student life and prepare for post-graduate professions. Similarly, school staff need to provide valuable, yet labor-intensive guidance to thousands of students every semester. It's a daunting and stressful task for both students and staff.
Striving to enrich the entire process while reducing costs, UCLA Anderson envisioned a generative AI-assisted roadmap for students. The goal was to gain efficiency in the systems used by students and staff while enhancing the experience from registration to job placement. Providing optimal AI-generated recommendations to students on registration materials and course selections, electives matching, social club and group membership content, career management tools, alumni mentorship programs and other services is essential to student success. The business impact for UCLA would mean that leading job placement numbers could drive greater student demand and higher alumni development revenue.
Our vision was to enhance the student experience through an AI-assistant that would help them with their own personalized roadmap and journey through our MBA program, from on-boarding through graduation.
Technical challenge
Before undertaking a large-scale deployment, UCLA wanted to demonstrate success with a proof of concept (POC) for its automated, GenAI virtual roadmap assistant. They needed a technology partner to co-create a solution that would centralize and integrate structured and unstructured student resources that are dispersed across several different data platforms and spreadsheets.
The solution also needed to comply with UCLA's IS-3 security standards, handle large volumes of data while adhering to privacy measures, integrate with the school’s environment and rely on machine learning (ML) to adapt and refine as it grows to constantly improve upon student queries. Finally, the user interface needed to be simple and provide accurate guidance.
Our solution
Together, UCLA, Kyndryl and AWS collaborated to co-create a POC platform that centralized data infrastructure within an AWS environment. The Kyndryl team performed enhanced discovery with in-depth Kyndryl Vital workshops and focus groups attended by UCLA faculty, guidance counselors, students and IT stakeholders.
The feedback uncovered pain points such as low resource utilization, student information overload and high operational costs. Kyndryl used the results to design an optimal solution to address the needs.
Working closely with AWS, Kyndryl established governance controls within UCLA’s AWS environment to ensure compliance with the UC IS-3 security standards to protect and mask student data. This required analyzing and mapping data sources and their corresponding meta data elements to provide a common framework to enable data modeling and governance. Together, ecosystem partners identified content sources and common student queries to develop data models that would power machine learning and GenAI capabilities. Finally, a simplified user experience was developed to enable real-world testing and refinement of the solution.
The power of partnership
Kyndryl and AWS leveraged their partnership to deliver an innovative solution that exceeded UCLA's requirements and demonstrated the strength of their alliance. AWS services were integrated into the single GenAI solution including:
The foundation included standard AWS components: AWS Lambda, S3 and API Gateway to ensure processing, compliance, security and reliability. The backend data was combined by utilizing Amazon Redshift data warehouse repository and visualization of results in Amazon QuickSight. The GenAI solution was powered by Amazon Bedrock foundation models and agents. Kyndryl utilized Amazon SageMaker and Glue Data Catalog to index data sets, deploy and train the models.
Our future goal is to refine and update the foundational elements provided in the POC to create a production-ready and cost-effective model that will help us navigate through budget cuts and economic uncertainty.