Вакансии

Figure’s Property Analytics team is seeking a Senior Applied Scientist to join us in our effort to reinvent real estate lending. You will be working alongside engineering, product, and business leaders to tackle complex research and machine learning problems related to automated valuation models (AVMs), asset valuation, and process optimization.

Location: Москва Может ли AI-агент сам выдвинуть гипотезу, поставить эксперимент и улучшить обучение языковой модели? В GigaChat Pretrain мы исследуем автоматизацию этого цикла: от выбора идеи и изменения кода до запуска экспериментов, анализа результатов и следующей попытки. Агент может искать новую архитектуру, подбирать рецепт обучения или оптимизировать GPU-ядра. За три месяца вы построите и исследуете такой цикл на выбранном классе задач. Цель - добиться измеримого улучшения и понять, что помогает агенту находить сильные решения. Обязанности Строить цикл «гипотеза → код → эксперимент → анализ → следующая гипотеза». Разрабатывать инструменты для агентов, экспериментировать с контекстом и стратегиями поиска. Проверять найденные улучшения: качество моделей, скорость обучения и воспроизводимость результатов. Требования * Уверенное владение Python и PyTorch. * Знакомство с распределенным обучением. * Умение отлаживать код, ставить эксперименты и разбираться в результатах. **Будет плюсом** * Собственные проекты с AI-агентами. * Опыт обучения языковых моделей. * Опыт с CUDA, Triton или оптимизацией GPU-вычислений.

Location: Москва Как сделать языковую модель качественнее при том же бюджете вычислений? В GigaChat Pretrain мы исследуем альтернативы стандартному вниманию: Gated DeltaNet, Kimi Delta Attention, Mamba и другие линейные и рекуррентные блоки. За три месяца вы разберёте перспективные подходы, выберете базовую архитектуру и проверите собственное изменение. Это исследовательская задача с неопределённым результатом: важно не только найти улучшение, но и понять, какие идеи работают, где и почему. Обязанности * Разбирать свежие статьи и выбирать перспективные идеи для экспериментов. * Реализовывать и менять линейные и рекуррентные блоки языковых моделей. * Сравнивать качество, скорость и расход памяти при сопоставимом бюджете вычислений. * Проводить абляции и выяснять, какие изменения стоит масштабировать. Требования * Уверенное владение Python и PyTorch. * Знакомство с распределённым обучением. * Понимание глубокого обучения и устройства Transformer. * Умение читать статьи, формулировать гипотезы и проверять их экспериментально. **Будет плюсом** * Собственный исследовательский проект. * Опыт обучения языковых моделей, работы с линейным вниманием или state-space моделями. * Опыт с CUDA, Triton или оптимизацией GPU-вычислений.

Location: Москва Можно ли изменить архитектуру уже обученной LLM и получить сильную модель дешевле, чем при обучении с нуля? В GigaChat Pretrain мы исследуем перенос обученных моделей в новые архитектуры: из dense в MoE, с добавлением экспертов и слоёв, изменением ширины или механизма внимания. За три месяца вы выберете одно направление, реализуете способ переноса весов и проверите, насколько быстро модель восстанавливает и улучшает качество. Главная метрика — сколько вычислений потребуется, чтобы достичь качества такой же архитектуры, обученной с нуля. Обязанности * Реализовывать методы переноса весов и адаптации модели к новой архитектуре. * Подбирать рецепт дообучения и сравнивать его с обучением с нуля. * Проверять собственные гипотезы, проводить абляции и оценивать экономию вычислений. Требования Что мы ждём * Уверенное владение Python и PyTorch. * Знакомство с распределённым обучением. * Понимание глубокого обучения и устройства Transformer. * Умение читать статьи и воспроизводить их результаты. **Будет плюсом** * Собственный исследовательский проект. * Опыт обучения языковых моделей, работы с MoE или дистилляцией. * Опыт изменения архитектур и переноса предобученных весов.

Location: Москва Модели распознавания речи чаще всего обучают на максимизацию правдоподобия идеальной транскрипции, а оценивают по количеству ошибок в распознанном тексте. Вы будете учить модель напрямую на недифференцируемую награду — сначала на Word Error Rate (WER), потом на точность редких слов из пользовательского словаря. Имена, термины и англицизмы — главный сорт ошибок в диктовке и голосовом ассистенте, а общий WER на них почти не реагирует. Пользовательский словарь — третий по частоте запрос пользователей наших открытых моделей Обязанности * Реализовывать on-policy обучение ASR-моделей с WER-подобной наградой * Добавлять контекстный биасинг: список слов в условие модели и награду * Сводить рецепт и переносить на продовый чекпоинт: сколько шагов, какая смесь, с какого момента деградирует общий WER Требования Уверенное владение python и pytorch Знакомство с распределенным обучением Знакомство с policy-gradient методами по статьям **Будет плюсом** * Опыт с verl, trl или любым другим RLHF-фреймворком * Опыт с vLLM или sglang * Любая практика с распознаванием речи, даже учебная

Location: Gurugram, Haryana, India NEOLYTIX Associate – Automation (Fresher) Intelligent Automation · Entry-Level · Start Your Career in AI & Automation EXPERIENCE Fresher (0–1 Yr) QUALIFICATION B.Tech / BCA / MCA LOCATION Gurugram (Onsite, 5-Day) SHIFT 1:30–10:30 PM (Flexible) About the Role Neolytix is a US-headquartered healthcare technology and services company with delivery centres in Chicago, Gurugram, and Manila, supporting 270+ provider organizations. As we move from a services company to a technology-enabled one, our Intelligent Automation team is building AI-powered automations that make healthcare operations faster and more accurate — and we’re hiring recent B.Tech / BCA / MCA graduates as Associate – Automation to grow that team. This is an entry-level role built for someone ready to learn and build. You’ll be trained to design, build, and test automations — using visual (no-code / low-code) workflow tools, AI models, and light scripting — for real business and revenue-cycle processes, under the guidance of senior automation engineers. No prior work experience is required. We’re looking for solid programming fundamentals, a logical, pattern-spotting mind, and genuine curiosity about AI and automation. What You’ll Do Learn to design, build, and test automations using visual workflow / RPA tools and AI models, under senior guidance. Support the Intelligent Automation team in building and testing automations for business and RCM workflows. Write and maintain simple scripts (Python) for data handling, integrations, and automation tasks. Help map manual processes step by step and identify where automation can save time or reduce errors. Document workflows, test cases, and process steps clearly. Monitor running automations, log exceptions, and flag issues to senior team members. Use AI tools (e.g., ChatGPT, Claude, Copilot) to prototype, draft, and test solutions faster. What You’ll Learn With structured training and mentorship, you’ll build genuine, in-demand skills in: Visual workflow automation and RPA (no-code and low-code platforms). Applied AI — prompt design and using AI models for real business use cases. Python scripting for automation, data handling, and integrations. Process analysis — how to break a manual workflow into automatable steps. How US healthcare revenue-cycle operations work, as the processes you’ll help automate. What You Bring Must-Have B.Tech, BCA, or MCA (recent graduate) with a genuine interest in automation and AI. Solid programming fundamentals — comfortable with at least one language (Python preferred); understands logic, loops, functions, and basic data structures. A logical, analytical, pattern-spotting mind — you can look at a process and see where the repetition is. Comfortable working with data in Excel / spreadsheets. Good spoken and written English for documentation and teamwork. A fast learner who takes ownership and sweats the details. Preferred (a bonus, not required) Academic or personal projects in Python, scripting, automation, or AI (a GitHub or portfolio helps). Exposure to any automation / RPA or workflow tool (UiPath, Power Automate, n8n, Zapier, or similar). Basic SQL; hands-on use of AI tools (ChatGPT / Claude / Copilot); interest in healthcare or US business processes. Eligibility Criteria Education: B.Tech, BCA, or MCA is mandatory and must be completed — candidates currently pursuing a degree, with backlogs, or an incomplete graduation are not eligible. Communication: Good spoken and written English — clear enough to document workflows and work within a team. Work mode & shift: Onsite at Gurugram, 5 days a week. Shift: 1:30 PM – 10:30 PM IST, with some flexibility as required. Availability: Immediate joiners or candidates able to start soon are preferred. Selection Process Application review (resume; projects, a GitHub, or a portfolio help). Technical / aptitude assessment (logic and basic coding). Interview with the Automation team, followed by an HR discussion. Why Join Neolytix Start in a High-Growth Field Begin your career in AI and automation — one of the most in-demand skill sets today. Learn & Build Structured training and mentorship from senior automation engineers, from day one. Real Work, Real Impact Build automations that run in a live healthcare business — not throwaway training exercises. Clear Growth Path A path to grow into Automation Analyst / Intelligent Automation Analyst as you develop. Questions? Reach out to Lakshay at Neolytix Talent Management   |   This document is confidential and for internal use only

Responsibilities: — Build and enhance biometrics, algorithms, and data-driven insights for the Mira App — Analyze hormone, cycle, app behavior, and user outcome data to identify patterns and product opportunities — Develop and validate models for ovulation prediction, cycle modeling, hormone interpretation, and personalized recommendations — Collaborate with Product, Design, Engineering, R&D, and Real-World Evidence Scientist teams to translate data into app features and evidence-backed insights — Support real-world evidence and clinical research projects through data extraction, exploratory analysis, modeling, and validation — Define model performance and validation frameworks, addressing uncertainty, subgroup performance, missing data, and edge cases — Improve data quality by identifying instrumentation gaps and recommending better data capture in the app — Document model logic, assumptions, limitations, and appropriate use cases clearly for both technical and non-technical teams — Work with Engineering to transition prototypes into production-ready app features or data pipelines Requirements: — Strong skills in Python and data science — Preferred experience with SQL and working with longitudinal, time-series, behavioral, health, wearable, physiological, or biomarker data — Experience with model interpretability and explainability methods such as SHAP and LIME — Strong judgment in model validation, including handling missing data, avoiding leakage, testing robustness, and evaluating performance across user segments — Ability to translate data into user-facing product features, metrics, recommendations, or algorithms — Experience building models, scores, or analytical frameworks that influenced product, clinical, or business decisions — Strong communication skills with both technical and non-technical stakeholders — High ownership, startup speed, and comfort working with ambiguity — Good scientific judgment, including distinguishing product insights from clinical or marketing claims — Experience moving machine learning or algorithmic features toward production Benefits: — Remote position — 40-hour workweek — Flexible working hours — Occasional overlap with the US team — Competitive salary with performance-based bonus system tied to OKRs — Fully remote environment supporting autonomy, ownership, and independence — Real opportunities for career growth, including the ability to shape, build, and develop teams as the company scales

Location: Москва Мы - команда инженеров, которые строят платформу для инференса и SFT opensource LLM. Используем продвинутые cutting edge движки для инференса, много экспериментируем и боремся за каждую дополнительную единицу в Token thoughput. Cейчас мы находимся в поиске Linux инженера, который возьмет на себя вопросы, связанные с разворачиванием, настройкой и тонким тюнингом core Linux для обеспечения оптимальной работы инференс движков. Обязанности ### Обязанности * настраивать, поддерживать и траблшутить GPU-инфраструктуру Блока * настраивать логирование, мониторинг и алертинг работоспособности всех компонентов инфраструктуры * коммуницировать со смежными подразделениями в части улучшений инфраструктурного ландшафта Блока — участвовать в оценках инфраструктуры под новые продукты и проекты. Требования ### Требования * продакшн опыт работы с GPU-инфраструктурой и инференсом SOTA LLM * опыт с ОС Linux: основные дистрибутивы, troubleshooting работы сервисов, оценка «здоровья» системы. * глубокое понимание этапов настройки ОС: сеть, диски, основные сервисы. * понимание устройства baremetal-сервера: CPU, RAM, сеть, хранение, шина PCI, топология GPU (NVLink/PCIe); ввод узлов в эксплуатацию, драйверы NVIDIA, CUDA * понимание сетевого уровня: протоколы TCP/IP, Ethernet, VLAN, MTU, TTL, DHCP, DNS, ARP, NTP. * опыт работы с инструментами мониторинга и логирования: Grafana, Elasticsearch, OpenTelemetry. * написание скриптов на bash, знание systemd * навыки программирования на Python. ### Будет плюсом * опыт работы с высокоскоростными interconnect-технологиями GPU-кластеров( RDMA/RoCe/Infiniband) и понимание работы NCCL. Условия ### **Условия** * гибридный формат работы (2-3 дня в офисе) из г. Москва * годовой бонус и ежегодный пересмотр * расширенный ДМС с первого дня + стоматологию и льготное страхование для семьи * корпоративный университет Сбера, внутреннюю образовательную платформу, участие в IT-конференциях * офис на Кутузовской с зонами отдыха и спортзалом * льготную ипотеку в Сбере, корпоративную пенсионную программу, подписку СберПрайм с возможностью совместного использования на трёх близких, скидки от партнеров и сервисов группы компаний.

​​​​​​ анализ данных и проверка гипотез для решения бизнес-задач; разработка и валидация ML-моделей (регрессия, классификация, кластеризация, ансамбли) feature engineering: подготовка и проектирование признаков;

Location: Paris; Amsterdam; London; Lausanne Workplace: Hybrid ABOUT MISTRAL Mistral provides full-stack AI solutions: from frontier models to developer tools, applications, and compute. We partner with enterprises tackling the hardest problems across high-stakes industries like finance, manufacturing, defense, healthcare, and the public sector, co-creating customized AI systems that they can run on their terms. We are a dynamic, collaborative team passionate about AI and its potential to transform society. Our diverse workforce thrives in competitive environments and is committed to driving innovation. Our teams are distributed between Europe, North America, Asia and the Middle East. We are creative, low-ego and team-spirited. THE ROLE Mistral AI is seeking a Lead / Staff Applied Scientist to help lead our robotics research and industrial collaborations in the Applied Science team. You will set the technical direction for robotics, and design, train, and deploy SOTA robotics models (VLM, VLA models for navigation and manipulation) with our industrial partners. Working cross-functionally with both external and internal science, engineering, and product teams you will deliver high-impact AI solutions. We are made of researchers, engineers, product builders, ex-founders who love tackling the most complex real world problems in the industry. WHAT YOU WILL DO - Set the technical direction for robotics in Applied Science: which problems, models, data, and benchmarks we invest in. - Lead the technical execution of high-stakes robotics projects with customers and partners, from scoping to deployment. - Build and mentor a team of robotics Applied Scientists and Research Engineers. - Run pre-training, post-training and deploy state of the art models on clusters with thousands of GPUs. - Own the data strategy: generate and curate datasets (image-text, video-text, teleoperation, simulation), and build evaluators/benchmarks for perception, navigation and manipulation tasks. - Develop the necessary tools and frameworks to facilitate data generation, model training, evaluation and deployment. - Collaborate with cross-functional teams to tackle complex use cases. - Manage research projects and communications with client research teams, and engage senior stakeholders. ABOUT YOU - Fluent in English with excellent communication skills — at ease explaining complex technical concepts to both technical and non-technical audiences. - Experience leading a technical AI, science or engineering team. Balancing technical IC work with team leadership and stakeholder management. - Expert with PyTorch or JAX. - Deep experience with Vision Language Models (VLM), Vision-Language-Action (VLA) models, perception, navigation and manipulation, simulation tools (e.g. Isaac SIM), with a track record of shipping models on real robots. - Not afraid of contributing to a big codebase; finds their way around independently with little guidance. - Writes clean, readable, high-performance, fault-tolerant Python code. - Doesn’t need roadmaps: moves independently and navigates uncertainty with confidence. - Low-ego, collaborative and eager to learn. - Track record of success through personal projects, professional projects or academia. NICE TO HAVE - Ex-founder or builder - PhD / Master in a relevant field (Robotics, Machine Learning) — exceptional candidates from other backgrounds welcome. - Experience with sim-to-real transfer, world models, or large-scale robot data collection. - Experience working on large code bases or large machine learning models at scale. - Contributions to a large open-source or industry codebase. - Publications in top academic journals or conferences. - Enjoys improving existing code: typing, tests, CI pipelines. WHAT WE OFFER We offer a comprehensive benefits package designed to support your well-being, growth, and work-life balance. Benefits vary by country and may include healthcare coverage, parental leave, retirement plans, relocation support, wellness programs, meal and transportation allowances, and other location-specific perks. For the most up-to-date details on benefits available in your location, please refer to our Benefits page https://app.notion.com/p/mistralai/Benefits-at-Mistral-36e6ba59a7fe836b93dd01737fcc27ef?source=copy_link. PRIVACY POLICY Your privacy matters to us. You can learn more about how we handle your personal data in our Applicant Privacy Policy https://legal.mistral.ai/terms/applicant-privacy-policy.

Location: Paris; Amsterdam; London; Lausanne ABOUT MISTRAL Mistral provides full-stack AI solutions: from frontier models to developer tools, applications, and compute. We partner with enterprises tackling the hardest problems across high-stakes industries like finance, manufacturing, defense, healthcare, and the public sector, co-creating customized AI systems that they can run on their terms. We are a dynamic, collaborative team passionate about AI and its potential to transform society. Our diverse workforce thrives in competitive environments and is committed to driving innovation. Our teams are distributed between Europe, North America, Asia and the Middle East. We are creative, low-ego and team-spirited. THE ROLE Mistral AI is seeking Applied Scientists and Research Engineers focused on optimisation and operational research. You will post-train SOTA large language models to formulate and solve optimisation problems, and apply them to diverse use cases. Working cross-functionally with both external and internal science, engineering, and product teams you will deliver high-impact AI solutions. We are made of researchers, engineers, product builders, ex-founders who love tackling the most complex real world problems in the industry. WHAT YOU WILL DO - Run pre-training, post-training and deploy state of the art models on clusters with thousands of GPUs. - Generate and curate datasets for optimisation modelling (LP, MILP, constraint programming, scheduling, routing), and build evaluators and benchmarks for operational research tasks. - Use solvers as verifiers to build reward signals for reinforcement learning. - Develop the necessary tools and frameworks to facilitate data generation, model training, evaluation and deployment. - Collaborate with cross-functional teams to tackle complex use cases. - Manage research projects and communications with client research teams. ABOUT YOU - Fluent in English with excellent communication skills — at ease explaining complex technical concepts to both technical and non-technical audiences. - Expert with PyTorch or JAX. - Experience training or fine-tuning large language models for reasoning, code, or mathematics. - Strong background in operational research or mathematical optimisation, with hands-on experience using solvers (e.g. Gurobi, CPLEX, OR-Tools) and modelling languages (e.g. Pyomo, AMPL). - Not afraid of contributing to a big codebase; finds their way around independently with little guidance. - Writes clean, readable, high-performance, fault-tolerant Python code. - Doesn’t need roadmaps: moves independently and navigates uncertainty with confidence. - Low-ego, collaborative and eager to learn. - Track record of success through personal projects, professional projects or academia. NICE TO HAVE - Ex-founder or builder - PhD / Master in a relevant field (Operational Research, Applied Mathematics, Machine Learning) — exceptional candidates from other backgrounds welcome. - Experience with reinforcement learning using verifiable rewards. - Experience with LLM-based optimisation modelling (e.g. NL4Opt-style problem formulation). - Industry experience in supply chain, logistics, energy, or scheduling. - Experience working on large code bases or large machine learning models at scale. - Contributions to a large open-source or industry codebase. - Publications in top academic journals or conferences. - Enjoys improving existing code: typing, tests, CI pipelines. WHAT WE OFFER We offer a comprehensive benefits package designed to support your well-being, growth, and work-life balance. Benefits vary by country and may include healthcare coverage, parental leave, retirement plans, relocation support, wellness programs, meal and transportation allowances, and other location-specific perks. For the most up-to-date details on benefits available in your location, please refer to our Benefits page https://app.notion.com/p/mistralai/Benefits-at-Mistral-36e6ba59a7fe836b93dd01737fcc27ef?source=copy_link. PRIVACY POLICY Your privacy matters to us. You can learn more about how we handle your personal data in our Applicant Privacy Policy https://legal.mistral.ai/terms/applicant-privacy-policy.

Location: Paris; Amsterdam; London; Lausanne Workplace: Hybrid ABOUT MISTRAL Mistral provides full-stack AI solutions: from frontier models to developer tools, applications, and compute. We partner with enterprises tackling the hardest problems across high-stakes industries like finance, manufacturing, defense, healthcare, and the public sector, co-creating customized AI systems that they can run on their terms. We are a dynamic, collaborative team passionate about AI and its potential to transform society. Our diverse workforce thrives in competitive environments and is committed to driving innovation. Our teams are distributed between Europe, North America, Asia and the Middle East. We are creative, low-ego and team-spirited. THE ROLE Mistral AI is seeking Applied Scientists and Research Engineers focused on tabular data. You will design, train, and deploy SOTA models that understand, query, and reason over structured data, and apply them to diverse use cases. Working cross-functionally with both external and internal science, engineering, and product teams you will deliver high-impact AI solutions. We are made of researchers, engineers, product builders, ex-founders who love tackling the most complex real world problems in the industry. WHAT YOU WILL DO - Run pre-training, post-training and deploy state of the art models on clusters with thousands of GPUs. - Generate and curate datasets of tables, spreadsheets and databases, and build evaluators and benchmarks for table understanding, text-to-SQL, and prediction on tabular data. - Develop the necessary tools and frameworks to facilitate data generation, model training, evaluation and deployment. - Collaborate with cross-functional teams to tackle complex use cases. - Manage research projects and communications with client research teams. ABOUT YOU - Fluent in English with excellent communication skills — at ease explaining complex technical concepts to both technical and non-technical audiences. - Expert with PyTorch or JAX. - Experience training or fine-tuning large language models, or tabular foundation models (e.g. TabPFN). - Strong background in machine learning on tabular data, including classical baselines (e.g. gradient boosting), and comfortable with SQL, pandas, and messy real-world data. - Not afraid of contributing to a big codebase; finds their way around independently with little guidance. - Writes clean, readable, high-performance, fault-tolerant Python code. - Doesn’t need roadmaps: moves independently and navigates uncertainty with confidence. - Low-ego, collaborative and eager to learn. - Track record of success through personal projects, professional projects or academia. NICE TO HAVE - Ex-founder or builder - PhD / Master in a relevant field (Computer Science, Statistics, Machine Learning) — exceptional candidates from other backgrounds welcome. - Experience with text-to-SQL, table question answering, or spreadsheet agents. - Experience with reinforcement learning using verifiable rewards (e.g. query execution). - Experience working on large code bases or large machine learning models at scale. - Contributions to a large open-source or industry codebase. - Publications in top academic journals or conferences. - Enjoys improving existing code: typing, tests, CI pipelines. WHAT WE OFFER We offer a comprehensive benefits package designed to support your well-being, growth, and work-life balance. Benefits vary by country and may include healthcare coverage, parental leave, retirement plans, relocation support, wellness programs, meal and transportation allowances, and other location-specific perks. For the most up-to-date details on benefits available in your location, please refer to our Benefits page https://app.notion.com/p/mistralai/Benefits-at-Mistral-36e6ba59a7fe836b93dd01737fcc27ef?source=copy_link. PRIVACY POLICY Your privacy matters to us. You can learn more about how we handle your personal data in our Applicant Privacy Policy https://legal.mistral.ai/terms/applicant-privacy-policy.

Опыт разработки ML-решений для бизнеса от 3 лет; Опыт работы с CV на реальных проектах от 1 года;

Location: San Jose, California, United States Who We Are At OKX, we believe that the future will be reshaped by crypto, and ultimately contribute to every individual's freedom. OKX is a leading crypto exchange, and the developer of OKX Wallet, giving millions access to crypto trading and decentralized crypto applications (dApps). OKX is also a trusted brand by hundreds of large institutions seeking access to crypto markets. We are safe and reliable, backed by our Proof of Reserves. Across our multiple offices globally, we are united by our core principles: We Before Me , Do the Right Thing , and Get Things Done . These shared values drive our culture, shape our processes, and foster a friendly, rewarding, and diverse environment for every OK-er. About the Opportunity We are looking for an experienced engineer to lead the architecture and hands-on development of the ML and data infrastructure behind our risk and fraud prevention systems. You will help teams move models from experimentation into reliable production services, improve model execution performance, and build scalable data foundations for training, evaluation, and real-time decisions. What You’ll Be Doing Improve model inference latency, throughput, resource utilization, and cost. Profile the full prediction path —from feature retrieval and preprocessing to model execution — and deliver measurable improvements without unacceptable model-quality regressions. Build reusable infrastructure for model packaging, versioning, deployment, serving, monitoring, and rollback across cloud platforms and regions. Design and optimize offline data pipelines on Databricks, Apache Spark, and warehouse/lakehouse technologies, addressing data modeling, partitioning, joins, shuffles, skew, incremental processing, and cost. Connect offline and online feature pipelines, ensuring point-in-time correctness, consistent feature definitions, freshness, lineage, and reproducible backfills. Evolve reliable feature services and data integrations supporting risk models and decision systems. Set technical direction, mentor engineers, and lead delivery across ML, data engineering, cloud infrastructure, security, and product teams. What We Look For In You A record of building and operating large-scale distributed systems, ML infrastructure, or data platforms, with technical ownership across teams. Hands-on experience optimizing production model performance, supported by measurable improvements in latency, throughput, utilization, or cost. Deep understanding of Databricks and Spark, or comparable offline data platforms, including execution plans, distributed query optimization, storage layouts, data correctness, and workload cost management. Experience with the model lifecycle, including training/serving integration, validation, deployment, observability, and rollback. Strong Python, SQL, and at least one backend or systems language such as Java, Scala, Go, or C++. Ability to turn ambiguous requirements into clear architecture and deliver maintainable systems through hands-on engineering and technical leadership. Nice to Haves SageMaker, Alibaba Cloud PAI/EAS, or comparable ML platforms. Inference optimization using tools such as ONNX Runtime, TensorRT, or Triton. Delta Lake, MLflow, feature stores, Kafka, Flink, and low-latency data stores. Fraud detection or financial risk systems. Fluency in Mandarin Perks & Benefits Competitive total compensation package L&D programs for employees' growth and development Various team building programs and company events Wellness and meal allowances Comprehensive healthcare schemes for employees and dependants More that we love to tell you along the process! OKX Statement: OKX is committed to equal employment opportunities regardless of race, color, genetic information, creed, religion, sex, sexual orientation, gender identity, lawful alien status, national origin, age, marital status, and non-job related physical or mental disability, or protected veteran status. Pursuant to the San Francisco Fair Chance Ordinance, we will consider for employment qualified applicants with arrest and conviction records. The salary range for this position is $214,666 to 321,999 The salary offered depends on a variety of factors, including job-related knowledge, skills, experience, and market location. In addition to the salary, a performance bonus and long-term incentives may be provided as part of the compensation package, as well as a full range of medical, financial, and/or other benefits, dependent on the position offered. Applicants should apply via OKX internal or external careers site. #LI-SK1 #LI-HYBRID Notice: All official OKX vacancies are published on this website. While roles may appear on selected third-party platforms from time to time, information on other sites may be inaccurate or outdated. If in doubt, please apply directly through our official careers website. Information collected and processed as part of the recruitment process of any job application you choose to submit is subject to OKX 's Candidate Privacy Notice .

Location: Canada What The Role Is We're hiring a Staff MLE for the Discovery team to own recommendations and personalization across Babylist's consumer experience — the homepage feed, product recs, search, and the ML-powered systems that make registry building feel effortless. Babylist was built on editorial recommendations: products chosen by people with deep baby gear expertise. That editorial foundation is a big part of why millions of families trust us. We're now building ML-powered personalization on top of it, using one of the richest first-party datasets in parenting. We're early in this work, and we have a real mandate. A small Discovery team has initial retrieval and reranking models live on parts of the site and a steady cadence of A/B tests. We are looking for a Staff MLE who has seen personalization done well at scale and can set the technical direction for where we go next. Registry building is the heart of the Babylist product. Every parent builds a list of dozens of products, from swaddle to stroller, with real stakes (a friend or family member is going to buy these things, and a baby is going to use them). As a universal registry, this registry also typically spans many retailers. This makes registry building on Babylist one of the most interesting personalization problems in consumer e-commerce: latent intent, life-stage progression, multi-stakeholder gift dynamics, deep declarative signal in millions of completed registries, cross-retailer datapoints, and a user who genuinely wants help. If you want to join a mature ML org and tune models at the margins, this isn't the right role. If you've worked inside a strong recommendations team, learned what good looks like, and want to build from zero-to-one at a company earlier in the journey, read on. What You'll Own You'll be the technical lead for personalization on the Discovery team, working alongside a PM, Engineering Manager, Senior MLE and fullstack software engineers. You set where our models go over the next year or two, sequence the bets that get there, and stay deep in building. A few examples of the problems you’ll get to shape: Product recommendations across every surface. Product detail pages, add-next recommendations after someone adds to their registry, vibes/aesthetic representations of products and the shared models underneath them. Personalizing the Feed. Deciding which products, editorial content and social content each family sees, and in what order. Personalizing search. Bringing what we know about a family into search results, and helping decide how much search should share retrieval and ranking with recommendations. Making Addie (our AI registry assistant) better at recs. Working with the team behind Addie so its product suggestions draw on the same models and signals as the rest of the site. In practice, you will: Take a fuzzy business problem from first sketch to production model, and own whether it moved the metric. Make the modeling and architecture calls that span surfaces and are expensive to reverse. Design the evaluation, offline and online, that tells us a model is good before and after it ships. Know when a rule or a simple baseline is the right first step, and when it's time to pursue more advanced approaches. Partner with product, design and data to help shape what's worth building. Coach engineers on the team, and more broadly be an ML expert and resource to the wider engineering org. Who You Are You’ve shipped recommendation and/or personalization systems that reached real users at scale within a consumer product, and can point to the business impact of your work. You’ve done this within a team that did ML well — you know what good looks like, and are motivated to bring that to a company earlier in their ML journey. You bring: Expertise in recommender fundamentals: for example, candidate generation vs. ranking, offline vs. online evaluation, cold start solutions, and handling position and popularity bias. Fluency in the Python ML stack and comfort owning models through deployment and monitoring, including contributions to our application backend. An ability to take an ambiguous problem and start moving before anyone hands you the full picture. A builder's instinct for early-stage ML. You understand effort vs. effectiveness tradeoffs and can advocate for when a rule beats a model, or a model is “good enough” for early learnings. An outcome and user focused mindset. You measure yourself by business impact, and you can connect a model change to user experience & registry adds, conversion or revenue. AI-native daily practice. You're already using AI to move faster and improve your output, and you stay curious about what's coming next. Compensation We post real numbers. For a Canada-based Staff Engineer, the starting base salary range is 299,300 to 372,600 CAD plus a target annual bonus of 20 percent of base. That's total target cash of roughly 359,160 to 447,120 CAD. On top of that you get meaningful equity and a 401(k) match. Where you start in that range depends on your experience, and your pay grows from there with performance and scope. How We Build AI is the default here. Engineers run agentic sessions for most of the work, and a lot of the interesting engineering now lives in the scaffolding that makes the agents good: the eval harnesses, the curated context, custom review skills and fast CI. Agents also triage incidents and handle a big share of support. A human always owns the outcome. The architecture is intentionally simple: one Rails monolith, MySQL and few moving parts. That's deliberate. Simple infrastructure lets us move fast and lets AI reason about the whole system, so the hardest problems are the ones in front of customers. The Stack Core app: Rails, Packwerk, React, TypeScript, Sidekiq Mobile: iOS (Swift), Android (Kotlin) Data, search & events: MySQL, Snowflake, dbt, Airflow, Weaviate, WarpStream, Hex, Sigma Machine learning: deep learning, matrix factorization, retrieval & ranking, AWS SageMaker, MLflow AI & dev tooling: Claude Code, Devin, Linear, CodeRabbit, Warden, LangChain, Bedrock AgentCore, Maxim Infra & ops: AWS, Kubernetes, Terraform, GitHub Actions, Datadog, incident.io, LaunchDarkly Key integrations: Shopify (payments), Iterable (CRM) Why Babylist An engineer, expecting her first baby, couldn't find the registry she wanted. So she built it. That's how Babylist started, and it's still how we work: engineers solving problems for families. Becoming a parent is one of the biggest moments in a person's life. Millions reach it for the first time every year, making thousands of decisions and figuring it out as they go. That's who we build for, and we're a long way from done. Ten million people give gifts through Babylist every year. We did more than $750M in revenue in 2025, up 45 percent over the year before, and we've been profitable for eight years while staying independent. So you can take on a hard, multi-year problem without watching over your shoulder for the next round or the next correction. And the team is small, around 65 engineers, so what you ship stays visible and your scope stays wide. How We Work Remote-first across the US and Canada, and we have been for years. That's not changing. We trust you to own your time and your outcomes, and we get everyone in a room together twice a year. Teams are small, pods of three to five engineers, so nothing you ship disappears into a committee. You'll work shoulder to shoulder with product, design and data, and with the partners across the business who rely on what you ship. You'll also stay close to customers yourself: sitting in on user interviews, watching session recordings, riding along with support. Here that's part of the engineering job, on a regular basis. How We Hire Three rounds, usually two to three weeks start to finish. Recruiter conversation (30 minutes). Trade context: what you want next, what we're building and straight answers on comp, team and remote. Technical screen (1 hour). One round, no AI, language-agnostic. Reach for Google when you normally would. We just want to see how you reason from first principles. Final round (4 hours). Four one-hour interviews: system design, AI-assisted coding with the tools you'd actually use here, product sense and culture and values. You'll hear back either way within 24 to 48 hours. If your timeline is tight, tell us and we'll move faster. Benefits Company-paid medical and fully covered dental and vision A 401(k) match Generous paid parental leave for birthing and non-birthing parents, plus a gradual return-to-work program Winter Wonder Week, a paid company-wide week off at the end of the year A remote-work stipend Mental-health and wellness support A Few Things To Know We record and transcribe interviews to evaluate candidates, in line with applicable privacy laws. We expect you to use AI in your work and we welcome it in the process, but what you submit and say should reflect your own thinking. If you have a family member or close relationship with a Babylist employee, let your recruiter know. Official outreach only ever comes from an @babylist.com address

Location: United States What The Role Is We're hiring a Staff MLE for the Discovery team to own recommendations and personalization across Babylist's consumer experience — the homepage feed, product recs, search, and the ML-powered systems that make registry building feel effortless. Babylist was built on editorial recommendations: products chosen by people with deep baby gear expertise. That editorial foundation is a big part of why millions of families trust us. We're now building ML-powered personalization on top of it, using one of the richest first-party datasets in parenting. We're early in this work, and we have a real mandate. A small Discovery team has initial retrieval and reranking models live on parts of the site and a steady cadence of A/B tests. We are looking for a Staff MLE who has seen personalization done well at scale and can set the technical direction for where we go next. Registry building is the heart of the Babylist product. Every parent builds a list of dozens of products, from swaddle to stroller, with real stakes (a friend or family member is going to buy these things, and a baby is going to use them). As a universal registry, this registry also typically spans many retailers. This makes registry building on Babylist one of the most interesting personalization problems in consumer e-commerce: latent intent, life-stage progression, multi-stakeholder gift dynamics, deep declarative signal in millions of completed registries, cross-retailer datapoints, and a user who genuinely wants help. If you want to join a mature ML org and tune models at the margins, this isn't the right role. If you've worked inside a strong recommendations team, learned what good looks like, and want to build from zero-to-one at a company earlier in the journey, read on. What You'll Own You'll be the technical lead for personalization on the Discovery team, working alongside a PM, Engineering Manager, Senior MLE and fullstack software engineers. You set where our models go over the next year or two, sequence the bets that get there, and stay deep in building. A few examples of the problems you’ll get to shape: Product recommendations across every surface. Product detail pages, add-next recommendations after someone adds to their registry, vibes/aesthetic representations of products and the shared models underneath them. Personalizing the Feed. Deciding which products, editorial content and social content each family sees, and in what order. Personalizing search. Bringing what we know about a family into search results, and helping decide how much search should share retrieval and ranking with recommendations. Making Addie (our AI registry assistant) better at recs. Working with the team behind Addie so its product suggestions draw on the same models and signals as the rest of the site. In practice, you will: Take a fuzzy business problem from first sketch to production model, and own whether it moved the metric. Make the modeling and architecture calls that span surfaces and are expensive to reverse. Design the evaluation, offline and online, that tells us a model is good before and after it ships. Know when a rule or a simple baseline is the right first step, and when it's time to pursue more advanced approaches. Partner with product, design and data to help shape what's worth building. Coach engineers on the team, and more broadly be an ML expert and resource to the wider engineering org. Who You Are You’ve shipped recommendation and/or personalization systems that reached real users at scale within a consumer product, and can point to the business impact of your work. You’ve done this within a team that did ML well — you know what good looks like, and are motivated to bring that to a company earlier in their ML journey. You bring: Expertise in recommender fundamentals: for example, candidate generation vs. ranking, offline vs. online evaluation, cold start solutions, and handling position and popularity bias. Fluency in the Python ML stack and comfort owning models through deployment and monitoring, including contributions to our application backend. An ability to take an ambiguous problem and start moving before anyone hands you the full picture. A builder's instinct for early-stage ML. You understand effort vs. effectiveness tradeoffs and can advocate for when a rule beats a model, or a model is “good enough” for early learnings. An outcome and user focused mindset. You measure yourself by business impact, and you can connect a model change to user experience & registry adds, conversion or revenue. AI-native daily practice. You're already using AI to move faster and improve your output, and you stay curious about what's coming next. Compensation We post real numbers. For a US-based Staff Engineer, the starting base salary range is $233,500 to $290,700, plus a target annual bonus of 20 percent of base. That's total target cash of roughly $280,200 to $348,840. On top of that you get meaningful equity and a 401(k) match. Where you start in that range depends on your experience, and your pay grows from there with performance and scope. How We Build AI is the default here. Engineers run agentic sessions for most of the work, and a lot of the interesting engineering now lives in the scaffolding that makes the agents good: the eval harnesses, the curated context, custom review skills and fast CI. Agents also triage incidents and handle a big share of support. A human always owns the outcome. The architecture is intentionally simple: one Rails monolith, MySQL and few moving parts. That's deliberate. Simple infrastructure lets us move fast and lets AI reason about the whole system, so the hardest problems are the ones in front of customers. The Stack Core app: Rails, Packwerk, React, TypeScript, Sidekiq Mobile: iOS (Swift), Android (Kotlin) Data, search & events: MySQL, Snowflake, dbt, Airflow, Weaviate, WarpStream, Hex, Sigma Machine learning: deep learning, matrix factorization, retrieval & ranking, AWS SageMaker, MLflow AI & dev tooling: Claude Code, Devin, Linear, CodeRabbit, Warden, LangChain, Bedrock AgentCore, Maxim Infra & ops: AWS, Kubernetes, Terraform, GitHub Actions, Datadog, incident.io, LaunchDarkly Key integrations: Shopify (payments), Iterable (CRM) Why Babylist An engineer, expecting her first baby, couldn't find the registry she wanted. So she built it. That's how Babylist started, and it's still how we work: engineers solving problems for families. Becoming a parent is one of the biggest moments in a person's life. Millions reach it for the first time every year, making thousands of decisions and figuring it out as they go. That's who we build for, and we're a long way from done. Ten million people give gifts through Babylist every year. We did more than $750M in revenue in 2025, up 45 percent over the year before, and we've been profitable for eight years while staying independent. So you can take on a hard, multi-year problem without watching over your shoulder for the next round or the next correction. And the team is small, around 65 engineers, so what you ship stays visible and your scope stays wide. How We Work Remote-first across the US and Canada, and we have been for years. That's not changing. We trust you to own your time and your outcomes, and we get everyone in a room together twice a year. Teams are small, pods of three to five engineers, so nothing you ship disappears into a committee. You'll work shoulder to shoulder with product, design and data, and with the partners across the business who rely on what you ship. You'll also stay close to customers yourself: sitting in on user interviews, watching session recordings, riding along with support. Here that's part of the engineering job, on a regular basis. How We Hire Three rounds, usually two to three weeks start to finish. Recruiter conversation (30 minutes). Trade context: what you want next, what we're building and straight answers on comp, team and remote. Technical screen (1 hour). One round, no AI, language-agnostic. Reach for Google when you normally would. We just want to see how you reason from first principles. Final round (4 hours). Four one-hour interviews: system design, AI-assisted coding with the tools you'd actually use here, product sense and culture and values. You'll hear back either way within 24 to 48 hours. If your timeline is tight, tell us and we'll move faster. Benefits Company-paid medical and fully covered dental and vision A 401(k) match Generous paid parental leave for birthing and non-birthing parents, plus a gradual return-to-work program Winter Wonder Week, a paid company-wide week off at the end of the year A remote-work stipend Mental-health and wellness support A Few Things To Know We record and transcribe interviews to evaluate candidates, in line with applicable privacy laws. We expect you to use AI in your work and we welcome it in the process, but what you submit and say should reflect your own thinking. If you have a family member or close relationship with a Babylist employee, let your recruiter know. Official outreach only ever comes from an @babylist.com address

Задачи: — участие в формулировке ИИ-гипотез, подготовке данных и обучении моделей, включая LLM-решения и классический ML — экспериментирование с LLM, prompt-инжиниринг, RAG и агентные подходы — разработка прототипов ИИ-сервисов — оценка качества моделей по метрикам, проведение офлайн- и онлайн-экспериментов — ведение технической документации и артефактов экспериментов Требования: — опыт работы в DS/ML не менее 3 лет в реальных продуктах — уверенные знания Python и библиотек NumPy, pandas, scikit-learn, PyTorch и других — опыт решения NLP-задач не-LLM методами — опыт или готовность быстро освоить LLM (развёртывание, настройка, prompt-инжиниринг, RAG/агенты) — понимание работы генеративных моделей и методов их дообучения — навыки подготовки и анализа данных, построения и интерпретации метрик — знание принципов создания ML-сервисов в on-prem (REST API, Docker) — понимание ML-жизненного цикла — английский язык для чтения статей и документации

Location: Алматы Задачи Разрабатывать ML модели для широкого круга задач; Участвовать в разработке сервиса на основе ML на всех этапах: от сбора данных до выкатки в продакшн и дальнейшего мониторинга/сопровождения сервиса; При необходимости оптимизировать модели под HighLoad системы; Тесное взаимодействие со своей продуктовой командой; Работать с нейронными сетями и обучать PyTorch модели для production; Интегрировать языковые модели (LLM/SLM) в существующие сервисы компании. Требования Понимание теории вероятности, математической статистики, линейной алгебры; Знание “классического ML” на хорошем уровне (регрессия, методы кластеризации, деревья решений и их ансамбли); Знание основ Deep Learning в сфере CV и NLP (работа с разнородными данными: изображения, текст, табулярные данные); Владение Python (опыт использования Pytorch) и связанным с ним стэком для ML (pandas, sklearn, numpy); Знание Git и Docker. Будет плюсом Опыт разработки ML решений в одной из следующих областей: NLP, CV, Recommendation Systems; Опыт работы с DVC, MLFlow; Опыт работы с airflow, bigquery; Понимание kubernetes.