As part of the Gates Foundation's Human Centered Design group, I collaborated with the Knowledge Management department to develop Aurora, a powerful search engine designed to accelerate the discovery of investment information relating to international organizations. Initially conceived as a basic keyword search tool, Aurora evolved significantly through the integration of Azure AI. This enhancement enabled the platform to intelligently analyze information across foundation documents and databases, providing more comprehensive and relevant data discovery and search results. A key priority was to ensure Aurora's accessibility and adaptability for a global user base with diverse roles and technical expertise. This involved designing an intuitive and user-friendly interface that seamlessly integrates into existing workflows.
Building upon previous research and design explorations conducted by the Human Centered Design team, I consolidated existing findings into a comprehensive slide deck, capturing insights and learnings from past user experience work. This deck enabled the team to gain a shared understanding of past work, identify key learnings, and establish a common ground for moving forward. This resource has been frequently referenced by the team, significantly streamlining the project initiation process.
REVIEWED RESEARCH: These were all of the research documents that were reviewed and summarized .
Soon after joining the project, I led the research and design of what was essentially a prototype of the search experience that was to be used to support another group in providing data to Knowledge Management. This required a rapid turnaround, necessitating a quick iteration between design and development, including integrating with Airtable (a separate web application). The challenge was to create a flexible structure that could be expanded upon for the official application. I created a clickable prototype and conducted initial user testing to gather feedback and refine the design. This prototype was successfully used by the collaborating group, providing valuable insights into user needs and technical feasibility. The learnings from this iterative process were crucial in informing the design and development of the official application.
PROTOTYPE: This was an example of detailed organization information within the prototype.
Building upon the learnings from the prototype, I led the creation of a new design system for Aurora, expanding upon the Fluent 2 framework. I collaborated with developers to customize and extend Fluent 2, creating new components and style guides tailored to the specific business, user and data requirements. This new design system ensured consistency across the user interface, improved development efficiency, and ultimately enhanced the overall user experience. The design system continues to evolve, adapting to new features and user feedback.
DESIGN SYSTEM: This is an example of the customized design system based on Fluent 2.
With the design system in place, the development of Aurora progressed efficiently. Given the four distinct content types, each requiring unique search results at the initial launch, careful attention to detail was crucial. To ensure the information architecture and user experience were effective, I conducted a series of user research studies. This included a card sort exercise to validate the organization of filters for the search results and usability testing of the developed prototype. The user research yielded positive results, with participants consistently praising the search engine's intuitive interface.
CARD SORT: The terminology similarity matrix from the card sort, produced from Optimal Workshop.
As the project progressed, the rapid advancements in AI technology necessitated a faster-than-anticipated integration of AI search capabilities. This presented new design challenges, including adapting the user interface to effectively communicate the use of AI while ensuring a seamless and intuitive user experience. I led the design of the AI-powered search features, conducting user research to understand user expectations and concerns regarding AI. We focused on developing clear and concise explanations of how AI was being used to enhance search results, while also addressing potential challenges such as bias and transparency. Through iterative design and user testing, we were able to successfully integrate AI into the search experience.
MOCK UP: The current design of Aurora, with AI integration through filter suggestions.
Since the successful launch of Aurora, I have been actively involved in analyzing user feedback and identifying areas for improvement. Based on user feedback, we are planning to expand beyond basic search results, incorporating features such as interactive geographic maps and personalized data visualizations. I am excited to continue working on Aurora and explore new ways to leverage AI to empower users with even more powerful and insightful information.
WIREFRAME: A future direction for Aurora, with conversational AI and map integration.
