The Journey

The problem

While establishing my independent research group as a UKRI Future Leaders Fellow, I encountered a practical barrier: difficulty accessing appropriate compute, storage, data infrastructure, training, and bespoke expertise for biomedical data science. Conversations with peers revealed this was not an isolated experience but a shared challenge for biomedical researchers across institutions and career stages.

It quickly transpired that no coordinating mechanism existed to connect fragmented institutional resources — infrastructure, training, skills, and people — into something coherent, accessible, and cost-effective. At a time when research institutions, particularly universities, are under severe financial pressure, the duplication and inefficiency are increasingly untenable.

Key lesson: If you are facing an infrastructure and/or skills issue, consider whether this is specific to you or a broader structural problem requiring collective action.

Scoping and evidence gathering

Without formal training in data science infrastructure, legal matters and institutional governance, I first sought to understand the problem in more detail — perhaps there was a good reason for what appeared to be an inexplicable lack of coordination and shared investment. Being a scientist, I of course began to gather key evidence.

• Informal consultations with researchers across institutions to determine whether the problem was shared
• An evidence survey distributed to biomedical researchers, asking about current infrastructure, training provision, skills gaps, unmet needs, barriers, and capabilities
• Institutional capability snapshots gathered from representative organisations to understand what existed — in compute, storage, platforms, training, and specialist staffing — what was duplicated, and what was missing
• Literature and comparator review looking at how other countries and regions had approached similar coordination challenges in data science infrastructure, workforce development, and training

The survey results highlighted a challenge that extended well beyond compute. Storage, workflow access, expert support, FAIR readiness, licensing costs, training fragmentation, and uneven institutional support were all significant barriers. A striking finding was that three-quarters of respondents were working with mission-critical biomedical data without formal training in data science — yet training resources, where they existed, were invisible across institutional boundaries.

This evidence base became critical for influencing upwards and bringing strategic stakeholders on board to support what was now turning into a structured initiative to enact change.

Key lesson: Systematic data collection is crucial. No matter how junior your position may be, if you make a strong case supported by data, you gain traction.

The first roundtable

Due to time and funding constraints, it was not feasible to run a workshop soon after the initial evidence was gathered. Instead, working with HDR UK Scotland, we convened a 2-hour virtual roundtable with diverse stakeholders (often their first time in the same room!) to discuss the emerging evidence and to map out key barriers and potential solutions. To remain focused, the roundtable featured presentations and two strategic breakout sessions aimed at co-developing the scope of a future in-person workshop. I adopted a strategy unblocking approach (“ship building”) to focus the sessions on movement from action to change.

The roundtable achieved three things:

1. Outlined barriers and key priorities — participants confirmed the challenges from their own perspectives, spanning infrastructure fragmentation, training invisibility, workforce precarity, and duplicated costs
2. Surfaced existing efforts — revealed that multiple parallel initiatives existed across infrastructure, training, and career development but were not connected to each other
3. Identified appetite for action — there was clear demand for and interested in a more structured approach to coordinating resources, people, and skills across institutional boundaries

Critically, the roundtable also revealed the diversity of stakeholders needed: infrastructure providers, data governance specialists, funders, institutional leaders, researchers, training organisations, and career development specialists. No single group could solve this alone — and no single dimension (infrastructure, training, or people) could be addressed in isolation.

Key lesson: An initial virtual roundtable that brings together key stakeholders early on is both a key diagnostic tool and a powerful opportunity for coordination and co-design across disparate communities.

Navigating obstacles

This period involved significant practical challenges:

• Timeline constraints — I moved between institutions during this period which caused extra pressure as the workshop coordination had to be done alongside research and a challenging team transition
• Scope management — the problem kept expanding as new dimensions (governance, IP, careers, training coordination) became apparent. What had started as a question about infrastructure access was revealing itself as a challenge spanning people, skills, resources, and institutional culture
• Financial pressures — universities were entering a period of severe budget constraint, making the case for coordination more urgent but the environment for action more difficult

Key timelines and milestones began to slip due to unforeseen challenges. I realised that the delivery of the workshop would be compromised unless it was rescheduled. With support from the FLF DevNet Team, HDR UK Scotland and colleagues at the CRUK Scotland Institute in Glasgow, rescheduling was coordinated succesfully and all stakeholders informed. Transparency throughout this process was key. The workshop planning was once again back on track.

Key lesson: Expect the unexpected, leverage key support networks and remain flexible with respect to planning as well as execution.

Designing the workshop

Working with a professional facilitator, the workshop was designed with a specific philosophy: move the room from diagnosis to decisions. This meant:

• Not rediscovering the problem — evidence summaries were prepared in advance so the room started from a shared baseline
• Structured discussions — each session had a defined question and a required output format (not open-ended “what do you think?”)
• Decision points — participants were asked to vote, rank, commit, and name deliverables

Four target outcomes agreed before the event:

1. Agreement in principle on a shared or federated model for infrastructure, training, and people
2. Design principles for phase 1 — what should be standardised nationally and what should remain local
3. Funding direction — pump-prime vs. baseline vs. hybrid, with clarity on what funding is needed for
4. Working group remit and membership, including initial deliverables and convening arrangements

Materials produced for the workshop included: a one-page aims sheet, an evidence summary covering infrastructure, training, and workforce data, survey highlights from respondents across nine institutions, an attendee expertise map, a barriers-and-enablers matrix organised by pillar, pillar-specific scene-setting sheets covering infrastructure, governance, authentication, data ownership, and careers and training, and structured discussion templates with sharpened questions designed to produce decisions rather than open-ended discussion.

Key lesson: Workshop design must be deliberate and carefully orchestrated for effective solutions. If designed for discussion, you get discussion. If designed for decisions, you get decisions.

The workshop

The in-person workshop brought together 40 stakeholders spanning:

• Researchers and data scientists (PhD student through to professorial level)
• Infrastructure providers and HPC specialists
• National data bodies and research data platforms
• Funding organisations and government representatives
• Institutional representatives
• Industry representatives
• Research software engineers, data stewards, and facility managers
• Legal and data governance specialists

Six structured discussions ran through the day, each building on the previous:

1. Agreement in principle — do we support a shared or federated Scottish model? (Result: Yes, with conditions)
2. Design principles — what should be standardised nationally and what should remain local, across infrastructure, training, and workforce coordination?
3. Pillar deep-dives — detailed work packages for infrastructure, governance, authentication, data ownership and IP, and careers and training
4. Funding direction — pump-prime vs. baseline vs. hybrid, with clarity on what funding is needed for and who contributes
5. Working group remit — what should happen in the first six months, including initial membership and convening arrangements?
6. Scoping roadmap — what does the next 12 months look like across all five pillars?

Key lesson: The right people in the room, with the right evidence, and the right questions, can achieve in one day what some committees may take months to accomplish.

Post-workshop: from outputs to ownership

The workshop produced a rich evidence base and a prototype framework spanning infrastructure, governance, authentication, IP and training. Another significant outcome, for me personally, was a transfer of ownership.

Senior colleagues — including representatives from national data infrastructure and training bodies — offered to carry the initiative forward. This was a key transition from grassroot-like advocacy led by me to collective momentum. The initiative was now a shared commitment to coordinating resources, people, and skills across Scotland’s biomedical data science landscape.

Post-workshop actions included:

• Synthesis of all discussion outputs into a structured document covering all five pillars
• Circulation to all participants with a call for working group membership
• Identification of proposed convening organisations with the breadth to span infrastructure, governance, authentication, IP, and training
• Initial scoping of a 12-month roadmap with phased deliverables across all dimensions of the framework

Key lesson: Once you have made the problem visible, built the evidence, and created the conditions for collective action, you can transfer ownership to those experts ideally positioned to deliver on the solution.