Guest: Adit Madan (LinkedIn)
Company: Alluxio (Twitter)
Show: Let’s Talk

Many enterprises are well aware of how generative AI can be leveraged by data teams to increase productivity. Alluxio has introduced a new version of its data platform to specifically help companies that need to train and fine-tune LLMs or other advanced AI models. Alluxio Enterprise AI has been purposefully designed to serve the needs of specialized compute hardware but without requiring a dedicated hardware to run on.

In this episode of TFiR: Let’s Talk, Adit Madan, Director of Products at Alluxio, talks about the announcement of their new data platform for AI, Alluxio Enterprise AI. He also talks about how  Alluxio is helping enterprise customers who are training and fine-tuning models.

Key highlights from this video interview :

• Madan gives us an overview of Alluxio, explaining it is a software/defined solution in the data infrastructure stack. He explains how they bring efficiency and ease of access to data.
• Generative AI is a hot topic right now and Madan discusses how it is based on a specific type of deep learning technique. He talks about the two different types of models, transformer architecture and convolutional neural networks (CNN) and he explains how early adopters of Alluxio are training on LLMs on top of Alluxio, which is based on the new transformer architecture and different types of CNN.
• Madan feels that companies do need AI to be part of the core strategy, particularly taking into account recent advancements. He discusses the impact AI is having on organizations improving productivity.
• Alluxio is focusing on enterprise customers who are training and fine-tuning models and Madan shares an example of one of their early adopters, a content company. He explains how they are using Alluxio to drive user engagement.
• Madan discusses which teams they are targeting within organizations, such as data analysts, data scientists, and specialized AI and training teams.
• Skills shortage continues to be an issue and a lot of learning that needs to be done around adopting these AI technologies. Madan discusses how Alluxio is helping enterprises with the experimentation phase, enabling them to try out new models and see the impact they have without a deep understanding.
• Madan takes us through some of the core components of Alluxio Enterprise AI, including a new system architecture, and a distributed object story repository architecture. He discusses how the architecture has been specifically designed to serve the needs of specialized compute hardware without requiring a dedicated hardware to run on.
• Madan describes the positive response they have had from early adopters of the solution. He feels that enterprises are educated on the need for the technology in space.
• The company’s initial offering, Alluxio Enterprise has now been forked into Alluxio Enterprise Data and the company has introduced a new product, Alluxio Enterprise AI, which is based on the new architecture and specifically designed for the AI/ML workloads.

This summary was written by Emily Nicholls.