Чёткая ответственность за весь ML-цикл, работа с мультимодальными моделями и небольшая профильная команда делают вакансию сильной; при этом важно учитывать обязательный офис и возможный дефицит GPU-ресурсов.
Только офис
Ограничения GPU
Современный ML
Ответственность за цикл
AI в коммерции
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
Senior/Lead Machine Learning Engineer for Product Data (AI): Building and deploying machine learning systems for product matching, catalog enrichment, and AI quality agents across a 150-million-product catalog with an accent on multimodal models, evaluation, and production-scale serving. Focus on optimizing GPU inference, calibrating precision and recall, and designing agents that identify and fix catalog errors under strict latency and cost constraints.
Location: Stockholm; on-site
Salary: SEK 868,446–1,108,832 per year
Company
Klarna is building an everyday finance network serving over 120 million consumers across 26 countries, with product discovery, price comparison, and agentic commerce capabilities.
What you will do
Take machine learning models from training through production, owning data pipelines, training, evaluation, and serving.
Determine model placement and GPU scaling while setting latency and cost targets.
Develop and measure AI quality agents that identify and fix catalog errors using product catalog tools.
Monitor production changes and explain model decisions to backend, operations, and product colleagues.
Help build a multimodal foundation model for unified, high-volume catalog predictions.
Requirements
Experience taking machine learning models from training to production in a live system.
Hands-on deep learning experience.
Production-quality Python skills, including tests, code review, CI/CD, and containers.
Ability to rigorously measure model performance and select appropriate metrics and thresholds.
Experience working with large language models and reasoning about quality, cost, and latency trade-offs.
Clear communication with backend, operations, and product colleagues.
Based in Stockholm and able to work on-site.
Nice to have
GPU inference optimization with Triton, TensorRT, or a similar technology, or experience training deep learning models at scale.
Experience with multimodal image-and-text models, including pretraining, fine-tuning, embeddings, or similarity search.
Experience building LLM agents with tool use through MCP, pydantic-ai, or a similar framework.
Experience building an ML pipeline from scratch and selecting its orchestration, storage, and training stack.
Culture & Benefits
Full-time role in a small, high-ownership team of six data scientists and ML engineers.
Frequent production releases and hands-on ownership of AI-native catalog systems.
Collaboration with colleagues across Europe.
GPU capacity may be limited at times, and the role regularly addresses questions about model decisions.
Hiring process
Submit a CV in English.
Prioritize concrete examples of what was built, its impact, and its cost over comprehensive descriptions.
Будьте осторожны
Если работодатель просит войти в их систему, используя iCloud/Google, прислать код/пароль, запустить код/ПО, не делайте этого. Это мошенники. Обязательно жмите «Пожаловаться» или пишите в поддержку. Подробнее в гайде →