Senior Applied Scientist, Amazon

• The Science and Tech (SnT) team of EU STEP is looking for candidates who are looking to impact the world with their mathematical and data-driven skills
• PhD and experience in computer science, optimization and/or ML or related field
• Experience in building models for business application
• Experience in patents or publications at top-tier peer-reviewed conferences or journals
• Experience programming in Python, Java or related language
• Experience in any of the following areas: algorithms and data structures, parsing, numerical optimization, data mining, parallel and distributed computing, high-performance computing

Desirable

• Solve complex optimization and machine learning problems using scalable algorithmic techniques
• Design and develop efficient research prototypes that address real-world problems in the massive logistics network of Amazon
• Lead complex time-bound, long-term as well as ad-hoc analyses to assist decision making
• Communicate to leadership results from business analysis, strategies and tactics
• You will be brainstorming algorithmic approaches with team-mates to solve challenging problems for Amazon logistics operations
• You will be developing and testing prototype solutions with above algorithmic techniques
• You will be scavenging information from the sea of Amazon data to improve these solutions
• You will be meeting with other scientists, engineers, stakeholders and customers to enhance the solutions and get them adopted
• We are the End-to-End Supply Chain optimizers
• As the core research team, we grow Amazon's logistics business to support decision making in an increasingly complex ecosystem of a data-driven supply chain and e-commerce giant
• Our mathematical algorithms provide confidence in leadership to invest in programs of several hundreds millions euros every year
• Above all, we are having fun solving real-world problems, in real-world speed, while failing & learning along the way
• We use modular algorithmic designs in the domain of combinatorial optimization, solving complicated generalizations of core OR problems with the right level of decomposition, employing parallelization and approximation algorithms
• We use deep learning, bandits, and reinforcement learning to put data into the loop of decision making
• We like to learn new techniques to surprise business stakeholders by making possible what they cannot anticipate
• For this reason, we work closely with Amazon scholars and experts from Academic institutions
• We code our prototypes to be production-ready
• We prefer provably optimal solutions than heuristics, though we settle for heuristics when performance dictates it
• Overall, we appreciate the value of correct modeling