Recommendation System Architect/lead developer

We’re looking for an experienced engineer to help develop the recommendation system for our social media–style application. An initial architecture is already in place, but it needs a critical reassessment. You’ll evaluate what’s there, find its weaknesses and bottlenecks, and improve or redesign it so it’s functional, scalable and production-ready, in line with industry-standard design principles. What you’ll be doing Work runs in sprint cycles: Reassess and finalise the architecture. Audit the current recommendation approach and codebase for gaps in scalability, performance, maintainability and recommendation quality. Propose and agree on an improved architecture covering data pipelines, candidate generation, ranking and serving, and decide what to keep, refactor or replace. Implement. Build the improved design, refactoring existing components where needed and keeping disruption to the wider platform low. Test continuously. Test your components on your own throughout development, covering unit, integration and offline evaluation of recommendation quality. Benchmark the new design against the current baseline. Integrate. Connect the recommendation system to other parts of the platform, such as user profiles, content and engagement data, the feed and APIs, and make sure it performs reliably at scale. What we’re looking for Proven experience building recommendation or personalisation systems, ideally for feeds, social or content platforms A track record of evaluating and re-architecting existing systems, not just building new ones A solid grasp of recommendation approaches: collaborative filtering, content-based methods, embeddings and ranking models Experience designing scalable backend systems and data pipelines The ability to explain architectural trade-offs clearly and justify design decisions High standards for code quality, testing and documentation The ability to work independently while coordinating cleanly with other components and team members Nice to have Experience with real-time or near-real-time recommendations Familiarity with cold-start problems, A/B testing and engagement metrics Experience with vector databases or feature stores Experience migrating live systems with minimal downtime