4:00 – 4:05 p.m. — Announcements

Colin Rundel opened the meeting and invited Evan Levine to share updates from the Co-Lab.

  • Evan reported that registration for Spring Co-Lab Roots was open, featuring more than sixty workshops. Although the sessions are often seen as student-focused, staff and faculty are also welcome to participate as part of professional development. New offerings include Web Scraping, APIs, AI with Raspberry Pi, RAG to Riches, and Mastering Effective Prompts.
  • He also highlighted the Course Chatbot pilot, developed with Mark McCahill and two faculty members the previous fall. After a successful trial, the Co-Lab will expand the pilot to additional courses this semester. Evan emphasized that the chatbot operates entirely within Duke’s secure environment and uses only course materials for training. He invited faculty with digital content to contact Michael Faber about participation.

     

4:05 – 4:35 p.m. — OIT Service Desk and Academic User & Device Support (Paula Batton, Peter Haman, Rodney Cozart, and Ed Gomes)

What it is: An overview of Duke’s IT support ecosystem—how users engage with the Service Desk and AUDS teams, how tickets flow through the system, and how service strategies are evolving as units merge into OIT.

Why it matters: As Duke continues to operate in a hybrid environment, reliable and consistent IT support remains critical. This presentation reviewed performance metrics, user feedback, and process improvements designed to enhance responsiveness and efficiency.

 

Paula Batton introduced Peter Haman, Assistant Director of AUDS, and Rodney Cozart, Senior Manager for the Service Desk. She noted that both teams have undergone significant change, integrating IT staff from Trinity, the Nicholas School, Student Affairs, and the Libraries into a unified OIT structure.

Tracy Futhey provided context, explaining that integration with Trinity IT began three years earlier and had progressed smoothly. She encouraged a close look at how consolidation affects service quality and user experience.

Service Desk Operations. Rodney Cozart described the Service Desk’s 24×5 schedule—from Sunday afternoon through Friday evening—with after-hours on-call support for major issues. Alongside full-time staff, fifteen to twenty student technicians provide coverage.

A major improvement has been the AI chatbot in ServiceNow that triages incoming requests. Implemented eight months earlier, it uses OIT’s knowledge base to offer automated assistance and passes unresolved questions directly to analysts. Since launch, self-service resolution has increased from 15 to 50 percent without additional development costs.

Rodney also noted the addition of an AUDS triage role, enabling real-time transfers from the Service Desk to specialists when an issue requires deeper access. The change reduced callbacks and shortened resolution times.

He highlighted the Duke Life Laptop Program, which supports students with financial need through a pool of more than 175 loaner devices maintained at The Link. Staff image, repair, and redeploy machines in coordination with the Duke Technology Store.

Through outreach at events such as Grad Fair, Tech Week, and Move-In Day, OIT improved visibility among students. For the first time, the student technology survey showed no reports of students being unaware of The Link’s location.

AUDS Team Structure and Efficiency. Peter Haman explained that AUDS now operates through geographically based teams—East, West, Central, Downtown, and Science Drive/Labs/E-Print—allowing closer oversight and greater consistency. Natural attrition reduced staffing by seven positions while maintaining coverage.

Combining unit-based teams streamlined imaging, deployment, and surplus workflows and eliminated duplication. Device Engineering, led by John Robinson, now manages backend provisioning for all units.

Paula Batton added that certain departments with specialized environments, such as those relying heavily on Linux, continue to receive tailored support.

Performance and Satisfaction. Rodney reported that the Service Desk and AUDS handled more than 51,000 tickets in the past year. Most requests arrived by phone, followed by email and chat. First-contact resolution was highest for phone and chat, where technicians could troubleshoot live.

Satisfaction ratings remain strong—96 percent for desktop support and 87 percent for the Service Desk. Paula encouraged members to share any negative experiences so her team can follow up directly.

Security and New Tools. Peter announced the deployment of Nudge, an application prompting users to schedule system reboots to complete critical patches. The tool lets users choose a convenient reboot time instead of being interrupted during work or class.

He also previewed a new Kuali onboarding workflow that will automate NetID creation, access requests, and equipment setup for new hires.

Challenges and Improvements. Paula acknowledged that turnover remains a challenge, as the Service Desk serves as a launchpad for IT careers across Duke. OIT is developing baseline training modules and stronger cross-unit relationships to preserve consistency.

Ed Gomes added that Trinity IT and OIT continue to refine communication channels and responsibilities to improve efficiency, particularly in research and Linux environments.

Steffen Bass described cases in which faculty requests were misrouted to the general Service Desk rather than to Science Drive IT, creating delays for complex research needs such as data retention. Ed Gomes acknowledged the issue, confirmed corrective steps, and noted that certain cases must still be escalated to IT Security.

Mark Palmeri praised overall responsiveness and suggested simplifying identity verification for authenticated callers. Tracy Futhey responded that new verification procedures are being developed following recent social-engineering attempts and that Nick Tripp will brief ITAC soon.

John Board proposed expanding front-line training so basic Canvas questions can be resolved without referral. Paula and Ed agreed this would improve the user experience.

Colin Rundel said his department’s transition to OIT support has been positive but that Linux users would benefit from clustered expertise rather than strict geographic assignments. Ed confirmed that OIT is evaluating this approach.

Steffen Bass asked how chatbot success is measured. Rodney explained that metrics rely on user feedback: resolved sessions count only when the user reports success; otherwise, the case is escalated.

Tracy Futhey thanked the Service Desk and AUDS teams for their continued work and encouraged departments to provide ongoing feedback.

4:35 – 5:15 p.m. — Federated Clinical Analytics Platform (FCAP) (Drs. Susanna Naggie and Svati Shah)

What it is: An overview of the Federated Clinical Analytics Platform (FCAP)—a collaboration with nference that integrates de-identified clinical, genomic, and pathology data for biomedical research within Duke’s secure environment.

Why it matters: FCAP provides a unified, privacy-preserving resource for cross-disciplinary research, addressing long-standing barriers to clinical data access while protecting patient confidentiality.

Dr. Svati Shah and Dr. Susanna Naggie described FCAP as a major advancement in Duke’s research infrastructure. Built with nference, the system combines structured and unstructured electronic health-record data, digital pathology images, and genomic information within a secure Azure environment under Duke control.

All data remain inside Duke’s firewall; none are sold or transferred externally. FCAP supports collaboration and potential revenue sharing with third parties while preserving compliance and institutional ownership.

Data and Capabilities. The platform holds de-identified EHR records for millions of patients, more than a million digitized pathology slides, and growing volumes of imaging and genomic data. About 150,000 patients will be consented through the One Duke Geninitiative, which includes whole-genome and exome sequencing and biospecimen collection.

Built-in analytic tools support data exploration and machine learning. Data are refreshed monthly, with dates shifted for privacy but chronological order maintained.

Privacy and Access. Access operates on two levels:

  • Explore Mode provides aggregated data for feasibility studies and educational use without IRB approval.
  • Premium Mode grants row-level access within approved protocols and allows computation inside Azure.

Results can be exported only after review to ensure de-identification.

Prasad Kasibhatla asked about patient consent. Susanna Naggie explained that Duke Health’s privacy policy allows de-identified data use with an opt-out option and community oversight.

Mark Palmeri and David MacAlpine asked about potential use in student training. Svati Shah said the team welcomes opportunities for teaching data analytics and computational biology.

John Board compared FCAP to the PACE environment. Susanna Naggie noted that FCAP offers greater computational power and complements rather than replaces PACE.

Tracy Futhey asked about cloud-cost management. Susanna confirmed that users receive usage alerts and spending limits.

Scott Schmidler and Mark Palmeri raised questions about commercial use and risk management. Svati Shah explained that data never leave the protected environment and that transparency strengthens trust in research.