B
Senior Machine Learning Engineer I – ML Production (Python, Java)
About Us: At Booking.com, data drives our decisions. Technology is at our core. And innovation is everywhere. But our company is more than datasets, lines of code or A/B tests. We’re the thrill of the first night in a new place. The excite…
Зарплата не указанаНидерландыAmsterdamГибрид
About Us: At Booking.com, data drives our decisions. Technology is at our core. And innovation is everywhere. But our company is more than datasets, lines of code or A/B tests. We’re the thrill of the first night in a new place. The excitement of the next morning. The friends you encounter. The journeys you take. The sights you see. And the memories you make. Through our products, partners and people, we make it easier for everyone to experience the world.
About the team: The ML Production team in Booking.com’s AI Platform track operates RS, the central inference platform for production workloads. Serving ranking, recommendations, fraud, pricing, and personalization, RS supports deployment, routing, and monitoring across real-time, offline, cloud, and on-premise setups.
Focus areas:
Role Description: As a Senior Machine Learning Engineer I in the ML Production team, you will design, build and operate the platform capabilities that make machine learning inference reliable and efficient at Booking.com scale.
This is a production-focused role. You will work primarily on integrating models into robust services and improving the platform that runs them, rather than developing models in isolation. You will combine machine learning knowledge with strong backend, distributed-systems and platform-engineering skills.
You will work mainly with JVM-based services such as Java and Scala, Kubernetes, heterogeneous infrastructure spanning multiple cloud environments and on-premise systems, and our observability stack. You will own services end to end: from requirements and technical design through implementation, deployment, operations and continuous improvement.
As a senior engineer, you will lead technically significant work within the team and across partner teams. You will help set the technical direction for platform initiatives, make sound trade-offs between reliability, latency, cost and developer experience, and coach other engineers through design discussions, code reviews and incident follow-up.
Key Job Responsibilities And Duties
Benefits & Perks - Global Impact, Personal Relevance:
Booking.com’s Total Rewards Philosophy is not only about compensation but also about benefits. We offer a competitive compensation and benefits package, as well unique-to-Booking.com benefits which include:
Inclusion has been a core part of our company culture since day one. This ongoing journey starts with our very own employees, who represent over 140 nationalities and a wide range of ethnic and social backgrounds, genders and sexual orientations.
Take it from our Chief People Officer, Paulo Pisano: “At Booking.com, the diversity of our people doesn’t just build an outstanding workplace, it also creates a better and more inclusive travel experience for everyone. Inclusion is at the heart of everything we do. It’s a place where you can make your mark and have a real impact in travel and tech.”
We ensure that colleagues with disabilities are provided the adjustments and tools they need to participate in the job application and interview process, to perform crucial job functions, and to receive other benefits and privileges of employment.
Application Process:
Pre-Employment Screening
If your application is successful, your personal data may be used for a pre-employment screening check by a third party as permitted by applicable law. Depending on the vacancy and applicable law, a pre-employment screening may include employment history, education and other information (such as media information) that may be necessary for determining your qualifications and suitability for the position.
About the team: The ML Production team in Booking.com’s AI Platform track operates RS, the central inference platform for production workloads. Serving ranking, recommendations, fraud, pricing, and personalization, RS supports deployment, routing, and monitoring across real-time, offline, cloud, and on-premise setups.
Focus areas:
- Building reliable, low-latency, high-throughput ML inference services
- Supporting online, async, and batch predictions
- Optimizing system performance, cost efficiency, and resilience
- Enabling safe deployment, rollouts, and lifecycle management
- Streamlining model integration and ops for practitioners
- Advancing cross-environment platform capabilities with portable abstractions
Role Description: As a Senior Machine Learning Engineer I in the ML Production team, you will design, build and operate the platform capabilities that make machine learning inference reliable and efficient at Booking.com scale.
This is a production-focused role. You will work primarily on integrating models into robust services and improving the platform that runs them, rather than developing models in isolation. You will combine machine learning knowledge with strong backend, distributed-systems and platform-engineering skills.
You will work mainly with JVM-based services such as Java and Scala, Kubernetes, heterogeneous infrastructure spanning multiple cloud environments and on-premise systems, and our observability stack. You will own services end to end: from requirements and technical design through implementation, deployment, operations and continuous improvement.
As a senior engineer, you will lead technically significant work within the team and across partner teams. You will help set the technical direction for platform initiatives, make sound trade-offs between reliability, latency, cost and developer experience, and coach other engineers through design discussions, code reviews and incident follow-up.
Key Job Responsibilities And Duties
- Architect, build, and operate scalable, low-latency distributed ML inference services and prediction systems across online, async, and batch workloads.
- Manage production model deployments, model versioning, feature access, shadow/canary testing, and rollbacks.
- Optimize compute utilization (CPU, GPU, memory) and lead capacity planning to minimize operational costs.
- Operate multi-cloud and on-premise Kubernetes workloads with automated scaling, CI/CD, and IaC practices.
- Implement end-to-end telemetry (metrics, logs, traces, alerts), maintain SLIs/SLOs, and participate in on-call response to drive service resilience.
- Partner with ML Scientists and Engineers to productionize models, improve developer experience, and mentor junior team members.
- Experience deploying and operating ML models in production across backend, infrastructure, or SRE domains.
- Proficiency in Java, Scala, JVM languages, or Python with strong backend engineering skills.
- Proven background designing distributed systems (microservices, RPC, queues, streaming, caching, batch).
- Solid grasp of concurrency, resource management, and server-side profiling.
- Hands-on experience managing containerized Kubernetes workloads in production.
- Experience with cloud/hybrid infrastructure, core platform capabilities, IaC, and CI/CD.
- Observability experience using SLIs, SLOs, metrics, logs, dashboards, and alerts.
- Understanding of ML serving (real-time/batch inference, versioning, feature retrieval, monitoring).
- Familiarity with on-call, incident response, postmortems, and runbooks.
- Pragmatic, customer-focused mindset for designing reusable platform capabilities and communicating technical trade-offs.
- Strong cross-team collaboration and English communication skills.
- Experience with ML serving platforms, runtimes (Triton, ONNX Runtime, TF Serving, TorchServe), or gateways.
- Background in JVM tuning, low-latency systems, GPU scheduling, or SRE/platform engineering.
Benefits & Perks - Global Impact, Personal Relevance:
Booking.com’s Total Rewards Philosophy is not only about compensation but also about benefits. We offer a competitive compensation and benefits package, as well unique-to-Booking.com benefits which include:
- Annual paid time off and generous paid leave scheme including: parent, grandparent, bereavement, and care leave
- Hybrid working including flexible working arrangements, and up to 20 days per year working from abroad (home country)
- Industry leading product discounts - up to 1400 per year - for yourself, including automatic Genius Level 3 status and Booking.com wallet credit
Inclusion has been a core part of our company culture since day one. This ongoing journey starts with our very own employees, who represent over 140 nationalities and a wide range of ethnic and social backgrounds, genders and sexual orientations.
Take it from our Chief People Officer, Paulo Pisano: “At Booking.com, the diversity of our people doesn’t just build an outstanding workplace, it also creates a better and more inclusive travel experience for everyone. Inclusion is at the heart of everything we do. It’s a place where you can make your mark and have a real impact in travel and tech.”
We ensure that colleagues with disabilities are provided the adjustments and tools they need to participate in the job application and interview process, to perform crucial job functions, and to receive other benefits and privileges of employment.
Application Process:
- Let’s go places together: How we Hire
- This role does not come with relocation assistance.
Pre-Employment Screening
If your application is successful, your personal data may be used for a pre-employment screening check by a third party as permitted by applicable law. Depending on the vacancy and applicable law, a pre-employment screening may include employment history, education and other information (such as media information) that may be necessary for determining your qualifications and suitability for the position.
Похожие вакансии Data Science & ML
Сайты компаний
Зарплата не указанаРоссияМосква
Мы ищем ML-инженера, который будет развивать качество e-commerce рекламы, используя технологию full-stream атрибуции https://ads.vk.ru/help/general/ecomm/dyn_ret/fsa. Наша команда занимается end-to-end разработкой и поддержкой всего пайплайна рекомендаций от обработки данных до выкатки в продакшен и проведения А/B-тестов.
Задачи
- Улучшение базового качества рекомендаций
- Развитие высоконагруженных сервисов рекомендаций
- Проведение и анализ A/B-тестов
Требования
- Опыт работы ML-инженером от двух лет
- Знание алгоритмов машинного обучения и принципов построения рекомендательных систем
- Опыт промышленной разработки на Java
- Понимание принципов работы высоконагруженных систем
Будет плюсом
- Опыт разработки рекомендательных систем и моделей ранжирования
- Опыт проведения и анализа A/B-тестов
- Опыт работы с YTsaurus, CatBoost, Spring и Java
hh.ru
Зарплата не указанаРоссияМоскваУдалёнка
UNIREST – правообладатель бренда ROSTIC’S, того самого с незабываемыми блюдами из курицы и крутой командой.
Обязанности:
- Разрабатывать ML-модели полного цикла: от гипотезы и экспериментов до внедрения, мониторинга качества и переобучения
- Взаимодействовать с командами DevOps и Data Engineering для интеграции и вывода моделей в production
- Оценивать эффективность ML-решений и их влияние на бизнес-показатели, в том числе с помощью A/B-тестирования
- Создавать и поддерживать ML-пайплайны в Airflow
- Презентовать результаты и выводы бизнес-заказчикам
Требования:
- Уверенное владение Python и SQL.
- Практический опыт работы с библиотеками Pandas, NumPy, scikit-learn, CatBoost и/или LightGBM.
- Умение интерпретировать результаты моделирования и объяснять их бизнесу
- Опыт использования Apache Airflow для оркестрации пайплайнов и Git для контроля версий
- Понимание принципов A/B-тестирования и методов оценки бизнес-эффективности ML-моделей
Мы предлагаем:
- Конкурентную заработную плату по итогам собеседования
- Годовую премию по результатам работы
- Гибридный формат работы (м. Войковская)
- Официальное оформление по ТК РФ в аккредитованной IT-компании
- Компенсацию затрат на питание в формате доплат
- Страхование жизни и ДМС после испытательного срока
- Обучение в Корпоративном университете и доступ к библиотекам Lerna и МИФ
- Командные мероприятия и корпоративные скидки
Присоединяйся к нашей команде – это отличное место работы
– именно такой сертификат выдан на основании наилучших показателей вовлеченности и лояльности персонала по итогам опроса сотрудников UNIREST на платформе Happy Job и включения UNIREST в список сертифицированных компаний
hh.ru
от 24 000 000 UZS (≈ от 166k ₽)УзбекистанТашкентОфис
О компании и о роли
Syncall AI ищет Senior NLP Engineer, который будет отвечать за качество языковых и голосовых моделей для языков с малым количеством данных: от сбора и чистки датасетов до модели, работающей в продакшене.
Syncall делает два продукта на узбекском и русском: Voice Agent — голосового ИИ-оператора для звонков клиентам, и Analytics — речевую аналитику разговоров. Наши агенты принимают входящие звонки, напоминают о платежах и продают. С нами работают крупнейшие компании в банковской сфере, телекоме, страховании, финтехе и ритейле.
Вы станете ключевым человеком в ML-команде: будете выбирать подходы, ставить эксперименты и доводить LLM, STT и TTS до уровня, на котором им можно доверить реальные разговоры с клиентами.
Задачи
- Файнтюн LLM под узбекский язык: continued pretraining, SFT, LoRA/QLoRA, при необходимости DPO и другие методы alignment.
- Файнтюн и адаптация STT-моделей (Whisper, wav2vec2, Conformer и аналоги) под языки, акценты, диалекты и шумную телефонную речь.
- Файнтюн TTS-моделей (VITS, XTTS, F5-TTS и аналоги): естественность, произношение, скорость синтеза.
- Сбор, разметка и чистка данных: поиск открытых корпусов, синтетические данные, аугментации, работа с разметчиками.
- Построение системы оценки качества: WER/CER для STT, MOS и разборчивость для TTS, бенчмарки и human eval для LLM.
- Оптимизация инференса под продакшен: задержка, стоимость, квантизация, стриминг.
- Выбор архитектурных решений и наставничество для middle- и junior-инженеров.
Обязательные требования
- 5+ лет в ML/NLP, из них 2+ года с LLM и/или речевыми моделями.
- Подтверждённый опыт файнтюна LLM для низкоресурсных языков: вы можете показать результат (метрики до и после, публикацию, репозиторий, модель на Hugging Face или кейс из продакшена).
- Практический опыт файнтюна STT-моделей и TTS-моделей, понимание их метрик и типичных проблем.
- Уверенный Python и PyTorch; опыт с Hugging Face (transformers, datasets, PEFT), распределённым обучением (DeepSpeed, FSDP или аналогами).
- Умение работать в условиях нехватки данных: transfer learning, кросс-языковой перенос, синтетические данные, аугментации.
- Умение самостоятельно ставить эксперименты, честно оценивать результат и объяснять решения команде и бизнесу.
Будет плюсом
- Навыки MLOps: пайплайны обучения и деплоя, трекинг экспериментов (MLflow, W&B), версионирование данных и моделей, мониторинг качества в продакшене.
- Опыт сервинга моделей: vLLM, TensorRT-LLM, Triton, ONNX; квантизация и стриминговый инференс для голоса с низкой задержкой.
- Опыт с Docker, Kubernetes, облачными GPU.
- Опыт построения голосовых агентов (связка STT → LLM → TTS) и работы с телефонией.
- Знание узбекского языка.
- Публикации, open-source вклад или участие в сообществах по низкоресурсным языкам.
Условия
- Зарплата: от 24 млн сум, итоговая сумма обсуждается по результатам собеседования.
- Формат: фуллтайм в офисе.
- Доступ к GPU для обучения и экспериментов.
- Большое влияние на продукт: ваши модели напрямую определяют качество разговора с клиентом.
Как откликнуться
Пришлите резюме в Telegram и коротко опишите один проект по файнтюну LLM, STT или TTS для низкоресурсного языка: какие были данные, что сделали и какие метрики получили. Ссылки на GitHub, Hugging Face или статьи приветствуются.
HireSeeker собирает вакансии со всех площадок и присылает только релевантные. Бесплатно.