Photograph of a swan at the University of Warwick (Phil Reed).

This year marked my first time attending RSECon, after several years of following along remotely via the UK Carpentries community calls. Being at the conference physically brought a new level of energy and connection, allowing me to participate more fully, both as an attendee and as a contributor. I had the privilege of helping to run a workshop during the main event, and I also led a session during the training community satellite day; both directly tied to my Software Sustainability Institute Fellowship and ongoing work developing the DIRECT Framework. I also enjoyed early morning strolls in the green campus of the University of Warwick.

The conference was full of inspiring content and collaborative spirit. I left with new tools and insights to bring back to my team. A few highlights included learning about the WAVE accessibility testing tool, which we’re already exploring for some of our internal documentation, and a tutorial on using Retrieval-Augmented Generation (RAG) to reduce hallucinations in large language model (LLM) outputs, an issue that’s especially relevant as we continue integrating AI into our products and services.

One of the standout aspects of RSECon was how it brought together multiple overlapping communities. I had the opportunity to connect with people working on related initiatives, such as STEP-UP, which focuses on developing digital research technical professionals (dRTP) around the London region. These interactions helped surface common goals and opened doors to potential collaborations, particularly around training and professional development.

While the entire event was valuable, I’d like to focus the rest of this post on the training community satellite day, where I led a session that aimed to identify and share useful training resources for research software engineers. The session was a hybrid one, and we used Miro as a digital whiteboard to collect contributions from both in-person and remote participants. This turned out to be especially effective, many of the suggestions were links to online resources, so having a shared digital space streamlined the process and saved us the work of transcribing handwritten notes

That said, the session wasn’t without challenges. It was designed as a collaboration between three overlapping groups, all sharing a single slot. Balancing different goals and expectations in a short amount of time was tricky, but everyone made a genuine effort to engage, and the volume of contributions we received was a testament to the shared enthusiasm. Miro requires a degree of control and instruction to be effective and accessible, so I made sure to include alternatives throughout the event.

Our next step within the DIRECT Framework team is to review the content generated during the session and continue mapping suggested courses to the professional skills section of the framework. We are also planning to liaise with the other group leads to synthesise the session outcomes, ideally culminating in a shared report or other output that captures the full day's collective knowledge.

Attending RSECon has been both energising and grounding. It offered a rare chance to step back and think strategically about how we support training and community-building in research software engineering. I’m already looking forward to building on these conversations in the months ahead.