Operational Issues

Description

A significant part of my role involves making — or contributing to — decisions about which technologies to use, build, or extend, and understanding why one approach is more appropriate than another in a given institutional context. The most sustained example of this is my work with Scientia, a bespoke Virtual Learning Environment developed within the Department of Computing at Imperial College London. Scientia serves approximately 1,500 students and 100 staff each day. My involvement as a core contributor means writing production code and reasoning carefully about trade-offs: performance versus flexibility, openness versus security, institutional ownership versus dependence on third-party vendors.

One decision I have thought about in particular depth is the choice to maintain a bespoke VLE rather than rely entirely on a commercial platform such as Blackboard or Canvas. Commercial VLEs offer scale, vendor support, and a predictable development roadmap, but they are not always well-suited to a research-intensive computing department with specific pedagogical requirements. Scientia allows us to design features — such as the PhD progression tracking module I led — that would be difficult or impossible to configure in a generic platform. However, bespoke development carries its own constraints: it requires sustained developer capacity, creates institutional knowledge risk if key contributors leave, and means that security patching and accessibility compliance fall entirely on the internal team rather than a vendor.

I have also participated in College-level technology selection processes, including the evaluation of a new VLE and the selection of FeedbackFruits as a peer assessment tool. These processes exposed me to a different kind of trade-off analysis: weighing factors such as data residency requirements, integration complexity with existing systems, total cost of ownership, and the realistic capacity of academic staff to adopt new workflows. The FeedbackFruits process, in particular, required me to think carefully about whether a feature-rich tool would actually be used — and used well — by the academic colleagues it was intended to support.

Understanding Operational Issues

My technical work spans several interconnected systems, and I will describe the most educationally significant of them here. The first is the Answerbook platform, which I manage for the delivery of digital examinations. Each year, Answerbook supports more than 500 students sitting assessments across 30 or more modules. My responsibilities encompass reliability, security, and scalability: I must ensure that the platform does not fail during a live examination, that student work cannot be accessed or tampered with by unauthorised parties, and that the system can handle simultaneous peak load from hundreds of concurrent users. This has required me to develop robust deployment pipelines — I use GitLab CI/CD throughout my work — and to establish clear incident response procedures in collaboration with the Exams Team, CSG IT, and invigilators.

The second system is ImPaaS, a bespoke Platform as a Service that I have extended beyond the Department of Computing to support other faculties. Bioengineering has now used ImPaaS for two years to host more than 30 group projects involving teams of four students. Extending a platform to new departments requires careful attention to infrastructure provisioning, access control, and user support — none of which can be assumed to transfer automatically from one institutional context to another. The experience also required me to work diplomatically across departmental boundaries, building trust with colleagues who had no prior relationship with our team or our systems.

A third area is the Teaching database, which I maintain as the departmental system of record for student enrolment, progression, and assessment. The integrity of this database has direct consequences for students' academic records, which means that technical decisions about schema design, data validation, and backup procedures carry genuine ethical weight.

I actively maintain expertise in AI and large language models as they apply to EdTech, and I am supervising a final-year undergraduate project that analyses data from the Lambda Feedback automated feedback platform to explore whether student interaction patterns can predict academic performance.

In this section I describe a training session for managing grades in Blackboard VLE