Our in-house search engine needs a sharp, data-driven eye to make every query deliver the right mix of products, articles, and images. I want you to analyse how users currently interact with results drawn from both textual content and rich imagery, spot the mismatches, then tune the ranking logic until relevance lifts our conversion rate. What you’ll do • Audit the existing pipeline, from query parsing to ranking, on our custom platform. • Work with mixed data (product descriptions, metadata, and image features) to understand which signals matter most for purchase decisions. • Design or refine relevance metrics and an evaluation set we can reuse for ongoing benchmarking. • Implement ranking improvements—whether that means boosting rules in Elasticsearch/Solr-style stacks, adding vector similarity, or training a lightweight ML re-ranker in Python. • Run statistically sound A/B tests and present clear dashboards so we see the conversion impact. Acceptance criteria • A repeatable evaluation suite showing at least X % lift in nDCG or equivalent relevance metric agreed upfront. • Deployment-ready configuration/code with rollback notes. • Concise report summarising methodology, key findings, and next optimisation steps. If you live and breathe query–result matching across text and images and can prove gains in real business metrics, let’s talk.