AWS Public Sector Blog
How the University of Arizona’s KMap transforms research collaboration with AI-powered discovery on AWS
When a University of Arizona associate professor needed to build a cross-disciplinary research team of women researchers for a new initiative, identifying the right experts seemed daunting. With thousands of faculty members spread across hundreds of departments, what would have traditionally required weeks of searching outdated websites and disconnected databases was completed in minutes through Knowledge Map (KMap), an AI-powered research platform developed by the University of Arizona and built on Amazon Web Services (AWS).
KMap goes beyond just traditional search to help the university understand and mobilize its knowledge assets. The platform functions as the university’s institutional brain, continuously learning from every research activity and growing smarter over time. Through an interactive map visualizing the entire research ecosystem, KMap connects principal investigators building grant teams, students seeking professors, new faculty finding collaborators, industry partners connecting with experts, and administrators tracking research strengths—enabling users to discover unexpected connections that drive innovation.
Managing 7,000 researchers across disconnected systems
The University of Arizona operates as an R1 research institution, managing more than 7,000 researchers across over 20 colleges and more than 150 departments. With over one billion dollars in annual research activity, 2,700 innovation disclosures, 600 patents, and 135 startups, the university generates an enormous amount of research data.
But scale creates complexity. Principal investigators spent weeks assembling teams for grant proposals. Students struggled to find professors whose research aligned with their interests. New researchers couldn’t quickly identify potential collaborators or campus activities. University leadership couldn’t get a clear view of departmental research strengths and activities. Most importantly, information across different systems was often outdated or incomplete.
“We needed an additional system so that it can look at all the data, bring it together in one place, keep the data up to date, and deliver these values to the research community,” explained Dr. Iqbal Hossain, director of research data science at the University of Arizona, who leads the KMap project.
Building the business case around “who is working on what”
The University of Arizona positioned KMap around a simple but powerful concept: “Who is working on what?” This universal question ties into multiple problems that cost universities time and money—from missed grant opportunities to inefficient team formation.
With that focus, the project received funding through the university’s Technology and Research Initiative Fund. The objectives were clear: improve grant response capabilities and strengthen private sector engagement. Support came from multiple campus offices, including the Office of Research and Partnership, University Analytics and Institutional Research, and Tech Launch Arizona—reflecting the university-wide recognition that better research intelligence was essential.
“The fundamental question is: how do you find people with the right expertise and build teams faster?” explained Hossain. The university wanted KMap to become an effective matchmaker for research, connecting the right people at the right time for maximum impact across all user groups. Delivering on that vision required a technology foundation that could handle the scale and complexity of the university’s research data.
Migrating to AWS for scalability and security
Initially, the University of Arizona operated with a mixed environment of on-premises and cloud servers. The decision to migrate everything to AWS was driven by the need for a unified data strategy across disparate systems. With research data scattered across dozens of sources, the university needed infrastructure that could integrate this information while securely implementing large language models (LLMs) and maintaining control over sensitive data.
The migration addressed critical challenges in AI implementation, particularly around data privacy and computational requirements. The university needed to deploy LLMs without exposing sensitive research data to third-party vendors while also maintaining the computational power and flexibility to test and prototype new features. Equally important was establishing clear data governance frameworks to ensure responsible AI use—defining who could access what data, how AI models would be trained and validated, and how the system would maintain researcher privacy while enabling discovery.
KMap now runs entirely on AWS, using services including Amazon Simple Storage Service (Amazon S3), AWS Lambda, Amazon OpenSearch Service, Amazon API Gateway, Amazon Bedrock, and Amazon SageMaker. The seamless integration between these services is essential for managing the complex data processing and AI workflows required for the platform.
Integrating dozens of data sources with comprehensive AI capabilities
KMap aggregates data from dozens of internal and external sources, including HR systems, publication databases, grant records, patent filings, and external platforms like Google Scholar, and ORCID. The system processes this diverse data through sophisticated AI workflows to create comprehensive researcher profiles and enable natural language interactions.
The platform employs generative AI in several innovative ways. LLMs extract research interests from publications and grants, creating uniform profiles across all researchers. “We use AI to discover everybody’s research interests based on all kinds of research activities,” said Hossain. The system automatically generates grant news summaries and provides an LLM-powered editing interface where researchers can paste entire publication lists for automatic processing and organization.
Most significantly, KMap features a retrieval-augmented generation (RAG) system that answers complex questions by finding relevant information across multiple documents. This allows users to ask natural language questions about university research capabilities and receive comprehensive answers drawn from the entire data ecosystem.
Beyond individual searches, KMap’s interactive map visualization helps users understand the bigger picture of research activity across campus. Departments appear as optimized polygons positioned based on actual collaboration strength, allowing users to zoom from a university-wide view down to individual researcher profiles and uncover unexpected connections across disciplines.
Figure 1: A visualization from KMap. Each polygon represents a department, and the blue circles indicate the total research funding received by individual researchers. Larger circles correspond to higher award amounts, providing an at-a-glance view of research activity and funding concentration across campus.
Figure 2: Internal collaboration network within a department visualized in KMap. The size of the circle corresponds to the researcher’s H-Index. The connecting lines illustrate patterns of collaboration and co-authorship, allowing users to see how expertise clusters and knowledge flows within and across disciplines.
Delivering measurable impact through rapid expert identification
The system has proven its value across diverse research scenarios, streamlining academic processes and saving significant time for researchers and administrators.
In 2023, when a nitric acid spill occurred near Interstate 10, close to the university, the communications team used KMap to quickly identify nitric acid experts for critical media interviews—locating the right specialists in minutes during an emergency. According to user feedback, researchers describe KMap as saving “a week of manual digging” with a single search.
The platform’s adoption has been substantial. More than 41,000 users visit KMap annually, generating 1.2 million interactions as they search for collaborators, explore research areas, and discover expertise across campus. Notably, 76% of traffic comes from organic discovery—users finding and returning to the system on their own rather than through mandated use.
The impact extends beyond individual searches. The university has measured stronger collaboration networks across departments, with proposal writing activities, creating particularly robust connections between previously siloed research areas.
Enhancing the platform with predictive features
The University of Arizona continues expanding KMap’s AI-powered features with a focus on proactive assistance. Future developments include collaboration recommendations that automatically suggest potential partners when users log in, and document-based discovery, where students can upload statements of purpose to find relevant professors automatically.
The university is also developing enhanced capabilities for international students who face challenges in identifying faculty with funding and research alignment. These students often spend extensive time searching for the right advisors and may not even know if potential supervisors are still at the institution.
For other universities considering similar initiatives, the university emphasizes starting with clear data governance and building incrementally. “Every research university needs this kind of system,” said Hossain. “But you don’t need to solve everything in one day. You can bring all your important data in, mine it, establish governance, and start building a similar kind of application so that it serves the community.” The approach focuses on specific use cases, gathering user feedback, and expanding capabilities based on demonstrated value rather than feature complexity.
The future of university research intelligence
More than just a search tool, KMap is an institutional research intelligence system that transforms how universities understand and maximize their research capabilities. By acting as an effective matchmaker, the platform accelerates discovery and innovation across disciplines, enabling stronger grant proposals, unexpected collaborations, and more effective responses to emerging challenges.
The University of Arizona’s success demonstrates that with the right technology foundation and clear vision, institutions can transform how researchers discover expertise, build teams, and collaborate across departments through AI-powered discovery platforms. As organizations worldwide seek to maximize their knowledge assets and accelerate innovation, KMap offers a model for the future of research intelligence systems: one where AI continuously learns from every research activity to unlock unexpected connections and drive discovery across disciplines.
Learn how AWS helps institutions build, deploy, and scale AI solutions that address public sector needs. Contact AWS today.