Samsung Research aims to deliver an end-to-end, data-driven user experience. We help Samsung device users enjoy seamless and personalized services powered by integrated data platforms within the Samsung device ecosystem. We leverage knowledge graph technologies to accelerate and optimize the data integration process, enabling a comprehensive AI-driven user experience. Moving forward, we are expanding this data platform to accelerate the development of personalized multi-device experiences.

We investigate systematic methodologies for improving model performance through data-centric optimization rather than architectural innovation. In addition to automated detection and correction of label noise, dataset imbalance, and distributional inconsistencies, this approach integrates synthetic data generation techniques—such as generative modeling, simulation-based augmentation, and large language model–driven data synthesis—to enhance coverage of rare or underrepresented scenarios. We evaluate how uncertainty estimation and influence-based data valuation can guide targeted data curation and synthetic sample injection. The project further explores slice-based evaluation frameworks to identify systematic failure modes and iteratively refine datasets. Ultimately, the research aims to establish principled metrics for dataset quality and quantify how synthetic data contributes to robustness, fairness, and generalization in real-world deployment environments.

We analyze data to uncover hidden user needs and support robust decision making. Our framework gathers and processes data from a wide range of sources. By coordinating with multiple teams and divisions, we define objective and effective problem statements that lead to well-founded solutions. Through these domain- and data-centered approaches, we ensure a consistent and integrated user experience across the entire Samsung device ecosystem. We actively employ a variety of methodologies, including conventional algorithms, machine learning, statistical analysis, and natural language processing techniques, to enhance our analysis solutions. Beyond analysis, we offer a dynamic platform that enables evidence-based decision making, streamlines workflows, and delivers deeper insights into user behavior.

Personal health data is scattered across countless sources—from wearable devices and mobile apps to clinical systems and hospital records. We are advancing research to integrate and analyze these diverse datasets to generate actionable, meaningful health insights.

By leveraging state-of-the-art health AI, we process large-scale health data to enable continuous management of chronic conditions and comprehensive monitoring of both mental and physical well-being. Life logs, personal health records (PHRs), and genomics form core inputs to our technology, empowering truly personalized healthcare services.

Our focus lies in developing a world-class, end-to-end AI platform that seamlessly integrates physical hardware infrastructure with advanced software systems, covering the entire AI development lifecycle—from data processing to model training and inference. Built on cutting-edge cloud technologies, we design autonomous and scalable platform architectures, and develop core technologies for their reliable deployment and operation. In particular, we prioritize establishing a cost-effective and sustainable AI development environment by researching and applying key optimization technologies—such as inference acceleration, distributed training, resource scheduling, and infrastructure automation—to maximize the efficient use of resources such as GPUs. Our ultimate goal is to provide standardized, easily accessible AI platforms and infrastructure to all AI developers across Samsung, fostering a collaborative ecosystem where resources and expertise are shared. By enabling synergy among teams, we aim to accelerate innovation and enhance productivity throughout the company.