Machine Learning Engineering Manager - Personalisation

Deliveroo , London, United Kingdom - Deliveroo

What you will do

Lead and develop a high-performing team of around 12 Machine Learning Engineers, distributed across multiple pods and spread across London, Finland and Berlin, fostering a culture of technical excellence, ownership and rapid experimentation.
Own and evolve the ranking, recommendation and personalisation systems across ordering surfaces, helping customers find the right items quickly as our selection grows.
Set the technical vision for the team’s ML ecosystem, balancing near-term product goals with longer-term bets on areas like on areas across recommendations to agentic approaches, and intent-aware experiences that adapt dynamically to how customers browse and act.
Lead the unification of our recommendation models across brands (Deliveroo, Wolt and DoorDash) and markets, working out how to power personalisation at global scale while keeping it feeling hyperlocal, and how to balance reusable core models against surface-specific adaptations without ending up with a sprawl of models.

Deliveroo , London, United Kingdom - Deliveroo

Lead and develop a high-performing team of around 12 Machine Learning Engineers, distributed across multiple pods and spread across London, Finland and Berlin, fostering a culture of technical excellence, ownership and rapid experimentation.
Own and evolve the ranking, recommendation and personalisation systems across ordering surfaces, helping customers find the right items quickly as our selection grows.
Set the technical vision for the team’s ML ecosystem, balancing near-term product goals with longer-term bets on areas like on areas across recommendations to agentic approaches, and intent-aware experiences that adapt dynamically to how customers browse and act.
Lead the unification of our recommendation models across brands (Deliveroo, Wolt and DoorDash) and markets, working out how to power personalisation at global scale while keeping it feeling hyperlocal, and how to balance reusable core models against surface-specific adaptations without ending up with a sprawl of models.