swap - Lead ML Engineer (recommendation systems)

• Own the end-to-end ML lifecycle for recommendation and personalisation systems, from problem framing and data exploration through to deployment, evaluation, and iteration. • Design, build, and productionise models for style-aware recommendations, including item pairing, outfit generation, preference matching, and personalised discovery. • Develop approaches that combine conversational preference extraction (from our memory layer) with traditional behavioural signals and LLM-based world knowledge to power high-quality recommendations, particularly in cold-start and sparse-data scenarios. • Build and optimise the feature pipelines and serving infrastructure that power recommendations at scale, working closely with engineering. • Define and champion best practices for offline and online evaluation of recommendation quality, including metrics for relevance, diversity, novelty, and style coherence. • Collaborate closely with product, AI engineering, and design to shape how recommendations surface across the AI Storefront, from conversational flows to visual discovery experiences. • Explore and integrate signals from social media content and visual style to enrich user taste profiles and improve recommendation relevance. • Act as a senior technical reference point for recommendation and personalisation engineering at Swap, helping to set standards, review critical work, and guide teammates. • What We Would Like to See • Significant experience (typically 5+ years) in ML engineering or applied machine learning roles, with clear ownership of production recommendation or personalisation systems that drove meaningful business outcomes. • Strong hands-on skills in Python and relevant ML/deep learning frameworks (e.g. PyTorch, TensorFlow), plus solid software engineering practices (testing, version control, code review, CI/CD). • Proven track record building recommendation systems, with practical experience in techniques such as collaborative filtering, content-based methods, embedding models, sequence models, or graph-based approaches. • Experience with LLMs and a practical understanding of how to leverage them within recommendation pipelines, whether for feature enrichment, preference understanding, knowledge bootstrapping, or hybrid retrieval approaches. • Comfort working with fashion, style, or visual domains is a strong plus, particularly experience with visual embeddings, multimodal models, or taste/preference modelling. • Practical experience deploying and iterating on ML systems in production (model serving, monitoring, retraining strategies, working with APIs and microservices).