JupyterCon 2023

Sangwoo Shim

Sangwoo Shim is Chief Technology Officer (CTO) and co-founder at MakinaRocks. He focuses on developing Machine Learning infrastructure and overseeing Al projects and solutions for manufacturing.
Sangwoo has pursued his career in various quantitative areas including quantitative finance, equity trading, and data analysis in global companies. Upon completion of his Ph.D. in Chemical Physics at Harvard University, he joined Bank of America Merrill Lynch in New York as a quantitative analyst. He later worked as a quantitative portfolio manager at WorldQuant and Millennium Capital in Old Greenwich and Singapore. Prior to MakinaRocks, he worked in the Big Data Analysis Group at Samsung Electronics.
Sangwoo believes that his expertise in statistics, optimization, and software engineering will contribute to achieving MakinaRocks’s mission to innovate manufacturing through artificial intelligence.

The speaker's profile picture


From Jupyter to MLOps: Jupyter as a key integrator for MLOps
Sangwoo Shim

The MakinaRocks Link plugin is a powerful tool that empowers data scientists to create, manage, and execute complex directed acyclic graphs (DAGs) within the JupyterLab environment. With its caching mechanism, support for parallelism, and remote execution capabilities, Link provides an efficient and user-friendly solution for building and deploying data pipelines. Additionally, Link complies with JupyterLab semantics, making it an ideal development platform for MLOps tools such as MakinaRocks Runway. By using Link to develop and test DAGs, data scientists can ensure seamless integration with their operational MLOps environment.

MakinaRocks Runway provides a comprehensive solution for managing the model training and serving pipelines. It simplifies the process of retraining models with new datasets and parameters, allowing users to accomplish this with a single click. Runway tracks training parameters, pipeline source codes, and the environment used to run training for each registered model. Users can also easily construct and update HTTP APIs and real-time inferences for their models. Updating ML models within running inference services with retrained models is a straightforward process.

In this presentation, we will explore the key features of MakinaRocks Link and Runway and demonstrate how they can help data scientists streamline their workflows and improve their productivity when constructing MLOps loops. We will showcase real-world examples of DAGs built using Link and highlight the benefits of using it in conjunction with MLOps tools such as Runway. Attendees will learn how they can work together to take their data science projects from development to the operational environment.

Louis Armand 2