Walmart · posted · checked
Staff, Data Scientist
Power biPythonPytorchTableau$143k – $286kOn-site · United StatesStaff+
Who can apply
On-site · United States
Location: (USA) Crossman Excellence Building CA SUNNYVALE Home Office
Position Summary...
What you'll do...
As a Staff Data Scientist at Walmart, you will leverage advanced analytical techniques and machine learning to transform complex data into actionable business insights. This role requires designing, implementing, and scaling high-throughput supervised fine-tuning (SFT) pipelines to train foundation models to follow user prompts, perform ranking tasks, and optimize training hyperparameters such as learning rate schedules and related configurations.
You will collaborate across functions to identify and evaluate data sources, ensure data quality, and communicate findings through effective visualizations. Your contributions will drive innovation and operational excellence by applying rigorous statistical methods and scalable solutions that align with Walmart's strategic objectives and commitment to delivering customer value.
About the Team
Our team develops foundational models tailored for Walmart's e-commerce platform by leveraging large-scale user interactions to create rich representations of users, products, and search queries. These models power scalable search, advertising, recommendation, and personalization solutions.
Team members apply expertise in data science, machine learning, and software engineering to optimize model performance, training, and deployment. With a strong focus on data quality, advanced analytics, and impactful visualization, the team delivers actionable insights that influence key business outcomes. Cross-functional collaboration drives innovation and enables continuous improvement in data-driven decision-making within a dynamic technology environment.
What You'll Do
Identify and evaluate data sources aligned with business requirements and quality standards.
Design, implement, and scale high-throughput SFT pipelines to train foundation models for instruction following, ranking, and related tasks.
Manage high-quality dataset mixtures and optimize training hyperparameters, including learning rate schedules and other training configurations.
Drive model alignment toward helpfulness, accuracy, and safety through preference optimization techniques and reinforcement learning approaches.
Create clear, insightful visualizations to effectively communicate findings and recommendations to stakeholders.