Роль четко определена с акцентом на машинное обучение и продуктовую стратегию, но отсутствие информации о зарплате является недостатком.
Сильная компания
Четкие обязанности
Нет информации о зарплате
?Гибридная модель работы
Мэтч & Сопровод
Для мэтча с этой вакансией нужен Plus
Подключить Plus
Описание вакансии
Текст:
Hirify AI/Оригинал
TL;DR
Principal Product Manager, Rides Matching (Machine Learning): Evolving Bolt’s ride-matching system from heuristic dispatch toward a context-aware allocation engine powered by machine learning and reinforcement learning with an accent on real-time allocation, marketplace supply-demand dynamics, and large-scale experimentation. Focus on coordinating model productionisation, maintaining the existing system during migration, defining matching-pricing interfaces, and optimizing allocation KPIs across markets.
Location: Berlin, Germany; London, United Kingdom; or Tallinn, Estonia. Hybrid model with at least 12 monthly in-office days; the workplace is also described as on-site.
Company
Bolt is a global mobility platform providing ride-hailing, food delivery, scooters, e-bikes, and car-sharing services across 50 countries.
What you will do
Define the product strategy and roadmap for the ride-dispatching engine, sequencing machine learning and reinforcement learning investments with improvements to queue ranking, dispatch strategy, and retry governance.
Coordinate productionisation of machine learning and reinforcement learning models from offline training through shadow mode, A/B testing, incremental rollout, and live monitoring across multiple markets.
Own the performance and stability of the existing matching stack while delivering its replacement and incremental efficiency gains.
Define and maintain the interface between matching and pricing, including reliable low-latency completion probability and supply-density signals.
Design experiments, evaluate results, and communicate algorithmic trade-offs through recommendations, strategy documents, technical specifications, and experiment reviews.
Set priorities and resolve trade-offs within a cross-functional pod of product, engineering, data science, and specialist roles, owning matching KPIs.
Requirements
Experience shipping machine learning models into production systems serving real-time allocations at scale, including feature engineering, training, shadow deployment, monitoring, and iteration.
Experience with allocation, matching, dispatch, or equivalent real-time systems in a two-sided marketplace, logistics platform, or high-throughput environment.
Strong understanding of supply-demand dynamics and global-versus-local optimisation trade-offs.
Experience maintaining a live production system while building its replacement, with a record of keeping current performance stable during migration.
Experience coordinating cross-functional product pods and working across organisational boundaries, including negotiating interface contracts and mutual accountability with adjacent teams such as pricing or economics.
Ability to perform detailed statistical evaluation and communicate clearly with senior stakeholders and cross-functional peers.
Ability to produce high-quality written strategy documents, technical specifications, and experiment reviews.
Direct production exposure to reinforcement learning, or experience working with teams that deployed reinforcement learning systems, including understanding of reward design, exploration-exploitation trade-offs, and differences between simulation and live environments.
Culture & Benefits
Small, autonomous teams with substantial ownership and responsibility.
Work on mobility products used by more than 200 million customers and 4.5 million driver and courier partners.
Salary and stock options.
Physical and mental wellness benefits.
A one-month paid sabbatical after five years at Bolt.
Annual company events and smaller team gatherings.
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 →