Data Scientist

Навыки

• A/B-тесты
• Airflow
• XGBoost / LightGBM / CatBoost
• CPC / CPM / CTR
• Elasticsearch
• GMV / MRR
• Highload

О компании и продукте

• Mayflower is a technology company building highload products used by millions of people worldwide. Operating at the scale of one of the world's top-50 websites, we solve complex engineering challenges and create solutions that power real-time entertainment for a global audience.
• Now we look for a Data Scientist to join our ML team

Задачи

• Search & Retrieval
• Develop and improve retrieval pipelines for large-scale production search systems
• Work on candidate generation, query processing, matching, filtering, and retrieval strategies
• Improve search relevance, result coverage, and overall SERP quality
• Analyse failed searches, irrelevant results, zero-result queries, and other search-quality issues
• Explore lexical, semantic, behavioural, hybrid, and vector search approaches
• Ranking & Relevance
• Build, train, and optimise ranking models for search and recommendation systems
• Develop learning-to-rank solutions using behavioural, content-based, contextual, and real-time features
• Design ranking features based on clicks, conversions, popularity, freshness, availability, and user behaviour
• Evaluate ranking quality using Precision, Recall, NDCG, MAP, MRR, and related relevance metrics
• Optimise models for low-latency inference and investigate relevance degradation, bias, and feedback loops
• Recommendation Systems
• Develop recommendation models and candidate-generation strategies for personalised and non-personalised scenarios
• Build recall and ranking stages for multi-stage recommendation pipelines
• Work on related-item, complementary-item, next-action, and behavioural recommendation use cases
• Develop user, item, session, and contextual representations
• Balance relevance, diversity, novelty, coverage, and business constraints
• Experimentation & Evaluation
• Design and run offline and online experiments for search, ranking, and recommendation improvements
• Build evaluation frameworks that connect model quality with product and business outcomes
• Design and analyse A/B tests using CTR, conversion, engagement, retention, and revenue-related metrics
• Create reproducible pipelines for data preparation, model training, evaluation, and comparison
• Evaluate model robustness across traffic segments, query groups, user cohorts, and edge cases
• ML Pipelines & Collaboration

Как проходит отбор

• HR interview (40 min)
• Technical interview (1.5 hour)
• Final interview (45 min)

Паспорт вакансии

История публикации

Откуда что взялось

Отмечено то, что вывели мы. Без пометки — значение назвал работодатель.

Почему на этом месте в выдаче

Порядок выдачи объявлен контрактом: свежесть решает между днями, полнота и зарплата — внутри дня.