3
Universities coordinated
Cross-institution AI delivery — 3 universities, one shared backlog
Delivered a chatbot initiative across a 3-university research coalition with shared backlog, aligned definition of done, and structured release planning.
AI chatbot delivered on schedule across 3-university coalition
Additional institutional funding secured at mid-grant review
Clinical accessibility requirements met across all user-facing features
Reusable delivery model documented for future multi-institution research projects
Shared backlog and definition of done adopted as coalition standard
Executive Summary
The University of Alberta, in coalition with two partner universities, needed to deliver an AI chatbot for individuals with neurodevelopmental disabilities — a research product with real end users, regulatory considerations, and three separate institutional stakeholder groups with competing priorities.
I owned delivery from concept through launch: writing user stories and acceptance criteria, managing the backlog across three institutions, running Agile ceremonies for a cross-university team, and coordinating multi-release delivery against grant milestones.
The project delivered on schedule, secured additional institutional funding at mid-review, and produced a reusable delivery model for future research-coalition projects.
Business Context
Research software delivery differs from commercial SaaS: stakeholders are academics with domain expertise but little delivery experience; timelines are grant-driven; and "definition of done" is contested across institutional interests. The chatbot's end users — individuals with neurodevelopmental disabilities — meant the product needed clinical input, accessibility requirements, and careful scope discipline to remain deliverable within the grant window.
Problem & Constraints
- Three universities with separate institutional stakeholders, approval processes, and priorities
- Grant-driven timeline with fixed milestones tied to funding review checkpoints
- Team members distributed across institutions with no standing delivery process
- Definition of done was contested across clinical, academic, and technical stakeholders
- No shared project tooling; minimal prior experience with Agile in the research context
My Role & Ownership
I led delivery coordination across the 3-university coalition — running ceremonies, managing the backlog, writing acceptance criteria, and coordinating multi-release delivery against grant milestones.
What I owned
- Agile ceremony facilitation across distributed multi-university team
- User story and acceptance criteria writing across all backlog items
- Backlog management and sprint planning across three institutional stakeholder groups
- Definition of done negotiation and documentation across clinical, academic, and technical requirements
- Multi-release delivery coordination aligned to grant milestone checkpoints
- Stakeholder reporting and mid-review preparation that contributed to additional funding
Not in my scope
- —Clinical requirements and neurodevelopmental disability domain (clinical research partners)
- —AI/ML model selection and chatbot architecture (technical research team)
- —Academic research design and publication rights (lead university)
- —Each institution's internal approval processes
Key Decisions
- 01
Established a single definition of done agreed by all three institutions before sprint one — prevented mid-project disputes over release criteria.
- 02
Set scope boundaries per release explicitly: which features were in which grant milestone, documented and signed off, so no university could unilaterally expand scope.
- 03
Created a shared backlog in a neutral tool (rather than any single university's system) to remove institutional ownership friction.
- 04
Introduced bi-weekly stakeholder sync with rotating note-taking — gave each institution visibility without requiring synchronous presence from all parties.
- 05
Prioritized clinical input on user stories in sprint planning, treating accessibility and safety requirements as non-negotiable acceptance criteria.
Actions Taken
Ran discovery workshops with clinical, academic, and technical stakeholders to align on user needs and definition of done.
Built and maintained a shared backlog with user stories and acceptance criteria accessible to all three institutions.
Facilitated sprint planning, retrospectives, and demo sessions across distributed time zones.
Coordinated multi-release delivery against grant milestones: scoped each release, managed dependencies, tracked completion.
Wrote mid-review stakeholder report demonstrating delivery progress that contributed to additional funding approval.
Documented delivery model and retrospective outputs for institutional reuse.
Delivery System & Process Improvements
- Multi-institution definition-of-done template documented and shared
- Shared backlog structure adopted for follow-on research work
- Bi-weekly cross-institution sync format reused in subsequent coalitions
Key takeaway
Research coalition delivery is a stakeholder alignment problem first. Clear scope boundaries per release and a shared definition of done — agreed before sprint one — are what make multi-institution projects deliverable.