We’re continually inspired by our customers and the complex data challenges they’re overcoming to make the previously impossible possible. In this blog series, we’ll spotlight the stories of some of our newest customers and why they chose WEKA.

Your organization’s AI efforts are only as good as their ability to deliver the performance and capacity to train AI models effectively. As the generative AI market continues to evolve at breakneck speed, supporting lightning-fast data storage access and large-scale GPU clouds requires a high-performance, high-capacity data platform.

Headquartered in South Korea, Upstage is an AI company specializing in creating unique large language models (LLMs) and document AI. Upstage’s flagship Solar LLM product delivers the performance and cost efficiency required for larger, more resource-intensive AI models, and it’s available on-premises and in the AWS cloud via API integration with Amazon SageMaker Jumpstart.

Upstage needed a modern data solution to ensure the performance required for scaling AI model development and to drive further experimentation that can meet exponentially growing market demands.

And we were happy to say, “WEKA to the rescue!”

Why They’re Cool

Upstage is a pioneering AI company powered by its unique Depth-Up Scaling (DUS) technique – a state-of-the-art training model that is already in use at leading enterprises across many industries and is outperforming pre-trained models from leading tech giants. DUS was uniquely trained on 7B (that’s right, seven BILLION) open-source AI models and engineered to prevent data leaks, mitigate AI hallucinations and deliver a robust, highly performant solution for AI researchers and developers.

Why They Chose WEKA

According to Upstage, WEKA possesses the cutting-edge technologies needed to optimize GPU performance for generating the best possible AI training models. This eliminates bottlenecks between GPUs and storage and plays a critical role in Upstage’s ability to showcase highly performing LLMs in the AI marketplace.

WEKA also provides an easy-to-use GUI interface, efficient performance maintenance, and easy integration with various monitoring solutions, enabling seamless monitoring of performance metrics and simplifying the process.

“At Upstage, having GPU servers and AI storage systems for model development is paramount, and WEKA has proven to be an excellent partner to collaborate with,” said Sung Kim, CEO of Upstage. “Since our initial implementation, we have been highly satisfied with the WEKA Data Platform’s stability and performance at scale and have experienced zero critical failures.  Using the WEKA Data Platform, we are now preparing to introduce new highly performant LLMs to the world and anticipate its impact will help us grow well into the future.”

Watch this Space

Upstage is just getting started with WEKA. We look forward to continuing our work with this emerging AI powerhouse and helping them to drive the industry forward. Learn more about their LLMs and developer tools here: upstage.ai

Explore WEKA for GPU Acceleration