Re-imagining recruitment
The client
REED, one of the UK's leading job sites and one of Europe's largest job boards. In 1995 it became the first high street recruitment agency in the UK to establish a web presence, and has been building on that foundation ever since.
Opportunity
REED's scale and heritage were real advantages - and real constraints. As nimble new market entrants emerged without the weight of accumulated systems to manage, REED needed to evolve both its technology and its business model to stay competitive. The site had been developed continuously since launch, but continuous development on ageing foundations eventually produces a platform that can sustain what it already does while struggling to support anything genuinely new. A specific internal challenge was improving data flow between teams and the AI and Machine Learning function, where friction was slowing the delivery of new site features and measurable outcomes.
The harder problem was not whether to modernise - it was understanding what the existing architecture could and could not support before committing to a direction. Moving quickly in the wrong direction is more expensive than taking the time to establish the right one.
Approach
Rather than moving straight to a build, Elemental Concept began with technical architecture consultancy - establishing what the existing systems could and could not support, and what new foundations were actually required before any new capability could be delivered reliably. That diagnostic step was what made the subsequent technology decisions defensible rather than assumed.
From that foundation, Elemental Concept provided technical and data architecture consultancy across several interconnected workstreams. The assessment identified that a root-and-branch redesign of major customer-facing systems was required. This led to redesigning the existing monolith into distributed microservices using an event-driven architecture, and implementing a Kong Enterprise API Gateway. To address the AI/ML data flow problem directly, EC built a fully automated Kubeflow-based CI pipeline that collected production data, orchestrated model training, and delivered updated models ready for deployment in the main site.
Outcomes
REED gained a technical foundation capable of supporting new business models, new delivery channels, and more efficient candidate-to-role matching - built on an honest assessment of what the existing architecture could carry rather than an optimistic one.
