Description

We are looking for a student enrolled in a Japanese university to work on applying large language models (LLM) to customer recommendation problems. The role is for 1-3 months with possibility to extend.

For this fellowship, you will complete an LLM project on recommendation and customer targeting. This requires you to build and validate data pipelines, perform extensive data cleaning and exploration, train and evaluate your models in a robust manner, design and conduct A/B tests to validate model performance, and automate model inference on AWS infrastructure. A mentor Applied Scientist will work with you to define the objectives and scope of the project, as well as provide regular feedback and consultation on a weekly basis.

The ideal candidate has built and validated a machine learning model for natural language processing (NLP) and tabular data. They must have strong skills in Python and its data analysis packages e.g. Numpy, Pandas, Matplotlib/Seaborn/Plotnine, as well as commonly used LLM packages e.g. HuggingFace. They have excellent understanding of how the ML models they train work under the hood. In addition, they may have worked with AWS infrastructure, distributed computing frameworks e.g. PySpark/SparkSQL.

The fellowship will be carried out from our Tokyo office in Meguro together with the rest of the team. Amazon will provide the necessary IT equipment (laptop, etc.) for the duration of the fellowship, a salary, and a stipend to cover commuting expenses.

Key job responsibilities

• Build and validate data pipelines for training and evaluating the LLMs

• Extensively clean and explore the datasets

• Train and evaluate LLMs in a robust manner

• Design and conduct A/B tests to validate model performance

• Automate model inference on AWS infrastructure

A day in the life

• On days of your internship at Meguro office, you usually start at 9-10 am and finish around 6-7 pm. You can flexibly determine your working hours as long as you are present for the meetings that require your attendance.

• Once a week, you will have a consultation with your mentor to update progress, get feedback, and suggestions for the project. You may schedule additional meetings or directly reach out to the mentor as necessary.

• Your mentor will work with you to set project milestones; you plan and execute the tasks to reach them at your own pace.

About the team

Economics and Decision Science Team is an embedded science team in the JP marketing space. Our ML models in recommendation and customer targeting directly interact with customers, creating immediate and long-lasting impact to their experience on Amazon.

We are open to hiring candidates to work out of one of the following locations:

Tokyo, 13, JPN

Basic Qualifications

• Currently enrolled in a Master's or PhD program in Computer Science, Data Science, Statistics, or a related field.

• Proficiency in Python programming language.

• Familiarity with large language model development and libraries (HuggingFace, LlamaIndex, Pytorch).

Preferred Qualifications

• Experience with natural language processing (NLP) techniques and libraries (e.g., NLTK, spaCy).

• Knowledge of deep learning frameworks and architectures.

• Previous experience with synthetic data generation or data augmentation techniques.

• Familiarity with AWS cloud computing platforms.

Please check the website below for measures to eliminate unwanted second-hand smoking in each facility:

https://www.amazon.jobs/en/landing_pages/passivesmoking

就業の場所における受動喫煙を防止するための措置に関する事項については、下記リンク先をご覧ください。

https://www.amazon.jobs/jp/landing_pages/passivesmoking

The salary information can be provided individually prior to the 1st interview

賃金に関する条件は、１次面接の前に個別にご案内することができます