Machine Learning Engineer, Routing and Planning Science

Does the idea of creating technology solutions for delivering 11 Billion+ packages across the globe excite you? If yes, come join a fun-loving, diverse, and creative team at Amazon Last Mile! The vision of the team is "To create Earth’s safest, most adaptive, and efficient plans for Last Mile logistics". The Last Mile Delivery Technology team is instrumental in impacting customer satisfaction directly, by devising innovative ways to deliver packages quickly and cost-effectively to the customers, and at scale using Artificial Intelligence (AI), Machine Learning and Operations Research solutions. Last Mile Delivery Technology organization supports the design, planning and execution of last mile transportation for Amazon’s various parcel and grocery delivery programs. All these programs require a large number of decision support systems to operate at scale and serve our customers, spanning demand planning, jurisdiction planning, delivery channel and network design, capacity planning for on the road and under the roof at delivery stations, routing inputs and route optimization. While these decision support systems have thus far been approached through the lens of traditional optimization and machine learning, we are looking to re-envision this space and pursue Foundational AI research, to innovate and advance the state of these decision support systems. Specifically, we are looking to develop foundational models (including Large Language Models, Multimodal Language Models, Multimodal Models), and adaptations to serve last mile use cases. Beyond Amazon the work developed will spur new fundamental knowledge and innovation in the logistics space.

We are looking for candidates with knowledge and experience in designing and developing state-of-art AI solutions (including development and application of advanced deep learning algorithms, foundational models, multimodal generative AI solutions, reinforcement learning), to join our team. Demonstrated experience scaling and optimizing the training and inference of these models, building real world solutions, to meet the customer use cases is preferred. Proven track record of publishing in major AI conferences and demonstrated experience working in industry will be preferred.

Job Location : Bellevue WA or Austin TX.

Key job responsibilities
Candidate will contribute to the development of Foundational AI solutions, to serve the myriad of decision support use cases across last mile. Candidate will be responsible for developing and implementing new machine learning techniques, optimizing training and inference of models, implement scalable science and engineering solutions in support of last mile planning and operations. You will be working with a team of scientists and engineers, helping shape the strategic and tactical vision, and contribute hands on to foundational research efforts in last mile.

A day in the life
Sitting at the core of Amazon’s Last Mile business, Routing and Planning (RaP) invents technology solutions that deliver toward the RaP vision: to create earth’s safest, most adaptive, and efficient plans for Last Mile logistics. RaP serves a diverse set of external and internal customers. For Amazon End Customers and transporters, we build solutions to deliver 11B packages and services to arrive at the right place and time, while meeting the growing preference for speed and precision. For internal customers, we build and delivery routing solutions Worldwide adding up to a $1B USD, across multiple amazon businesses ranging from Amazon Logistics, Amazon X-tra Large to Same/Sub-Same Day Deliveries, Fresh and Groceries etc using multiple mode of transports. If the opportunity to innovate, create and deliver first-time solutions alongside world-class engineers and scientists e.g. building a new delivery foundational model from scratch utilizing data at an unprecedented scale in the industry excites you, this is the place for you.

About the team
The Last Mile (LM) Science organization serves two primary functions: (1) maintaining and enhancing the components which provide the essential intelligence in nearly every LM system, and (2) creating the science which will enable the continued growth of Amazon in the future. The systems used on a daily basis by the Last Mile to deliver packages from the station to the customer’s doorstep contains embedded models developed by the science team, or relies on inputs derived from these models. These models determine how efficiently Amazon can serve its customers and at what cost.