Additional Information
Job Description:
Department Overview
Artificial Intelligence Office is dedicated to leveraging data science and machine learning to provide solutions that enhance the various functions of Rakuten Travel. From product development to marketing, sales, advertising, and quality control, we use data-driven insights and predictive analytics to drive innovation and deliver measurable results.
Position:
Why We Hire
We are currently seeking talented ML engineers to join our team and take on the responsibility of optimizing Rakuten Travel's marketing activities. In a rapidly changing online travel industry, it is crucial that we provide our customers with a more personalized and tailored experience. As such, we are constantly on the lookout for enthusiastic data scientists who can help us deliver the best possible value to our customers. Additionally, we offer a supportive environment where we analyze user logs and run various experiments to continuously improve our products through iterative development.
Position Details
As a member of our team, you will work closely with business side to understand their challenges, collaborate with data engineers and front-end engineers, and deliver solutions that optimize Rakuten Travel's product offerings. Specifically, you will be responsible for projects such as item recommendation, user targeting, image optimization, and NLP tasks utilizing user reviews. In this role, you will have the following key responsibilities:
- Collaborate with business side to define project requirements.
- Develop detailed dev specifications.
- Build and deploy ML models via APIs.
- Design and execute experiments to validate solutions.
- Conduct AB tests and bandit optimizations to compare different approaches.
- Evaluate experimental results using statistical tests to measure impact and effectiveness.
Mandatory Qualifications:
(1) Possess a passion and curiosity for the fields of machine learning and data science.
(2) Hold a master's degree or higher in machine learning, mathematics, or a related field, or have equivalent practical experience. Additionally, have experience in building machine learning/deep learning models, including those related to image processing, natural language processing, and other related areas.
(2-1) [Guideline] Expected to have knowledge in one or more of the following areas:
(2-2) Machine Learning: Foundational knowledge such as machine learning models like discriminative and generative models, inference algorithms like stochastic gradient descent, and asymptotic theories like asymptotic universality and asymptotic efficiency, as well as one of the following:
(2-2-1) Counterfactual Machine Learning: Various estimators for off-policy learning. Sampling algorithms like Importance Sampling.
(2-2-2) Attention/Transformer: Ability to explain the mathematics of Attention and Transformer. Model merging and MOE. Ability to explain each training step of large language models.
(2-2-3) Optimal Transport: Understanding of basic concepts such as the Sinkhorn algorithm and Wasserstein distance.
(2-3) Mathematics: Linear algebra (diagonalization, Jordan canonical form), calculus (differentiation + Riemann integration), topology (topology, compactness, continuity), as well as one of the following:
(2-3-1) Analysis: Measurable functions, integrable functions, definition of probability spaces. Probability process theory.
(2-3-2) Geometry: Definitions of Riemannian manifolds, Lie groups, and statistical manifolds. Distances such as Gromov-Hausdorff distance.
(3) Have development experience using Python.
(3-1) [Guideline] Expected to have experience in one or more of the following:
(3-2) Development experience using PyTorch or TensorFlow.
(3-3) Execution of training using GPUs.
(3-4) Experience in having Python code reviewed by others.
(4) Possess expertise in experimental design for conducting A/B tests and similar experiments.
(4-1) [Guideline] Expected to have knowledge in one or more of the following:
(4-2) A/B testing based on statistical hypothesis testing. Minimum sample size design using non-central t-distribution.
(4-3) Bayesian A/B testing.
(4-4) Optimal arm decision algorithms.
(4-5) Interleaving algorithms.
(5) Able to work in a cross-functional team environment and collaborate effectively with other teams.
Rakuten · Tokyo, Japan