In Amazon EU Supply Chain & WW Operation Compliance, we optimize our multi-constraints supply chain end-to-end, learning from and influencing real time large-scale execution systems, by bringing operations research and machine learning together into production. We also automate the classification of the products on our catalogue within the different regulatory classes according to the IATA regulations.
With the computational developments and accessibility efforts made by Amazon Web Service, we can now leverage dedicated services like Amazon Bedrock and Sagemaker JumpStart  hosting leading foundation models (Anthropic, Meta..).

With the recent advancement in language modelling and transformers encoder/decoder based we want to leverage these techniques to automate manual tasks/provide support to humans in their daily tasks ranging from product classification, documents review and information extraction through custom AI assistants/models.
The students will structure and execute an experiment protocol, benchmarking different techniques available in this domain (e.g. prompt engineering, fine-tuning, knowledge grounding, RLHF) and evaluate them according to internal and external success metrics.
A labelled dataset inclusive of more than 1MM human-labelled products will be provided for fine-tuning purposes as well as datasets used in the literature to benchmark the models across different tasks.
The students will also test language and multimodal models (e.g. CLIP/BLIP/BLIP-2) leveraging raw images from the Amazon Detail Page and image embedding in the model pipeline.

The students will be provided with full access to an AWS account linked to EC2, S3, Sagemaker, Amazon Bedrock.
We see this as a great opportunity to familiarize the students with AWS technologies which will provide on hand experience and be part of an exciting and innovative research and development project to develop new solution in state of the art programming languages. This will provide fruitful on-hand experience to the students on highly demanded and value added skills (PyTorch, Python; familiarity with AWS technology) and it will also be an opportunity for them to review and dive into machine learning models for classification.

Apply to this project here