AI News sat down with Predibase CEO and co-founder Piero Molino during this year’s AI & Big Data Expo to discuss the importance of low-code in machine learning and the LLM (Large Language Model) trend. .
At its core, Predibase is a declarative machine learning platform intended to streamline the process of developing and deploying machine learning models. The company is on a mission to simplify and democratize machine learning, making it accessible to both professional organizations and developers new to the space.
The platform empowers organizations to leverage in-house expertise to cut development time from months to just days. Additionally, it caters to developers who want to integrate machine learning into their products but lack the expertise.
By using Predibase, developers avoid writing a lot of low-level machine learning code, instead writing a simple configuration file (known as a YAML file) containing just 10 lines specifying the data schema. can work with
Predibase is now generally available
During the show, Predibase announced the general availability of its platform.
One of the key capabilities of this platform is the ability to abstract the complexity of infrastructure provisioning. A user can seamlessly run training, deployment, and inference jobs on a single CPU machine, or he can scale up to a 1,000 GPU machine with just a few clicks. The platform also facilitates easy integration with various data sources such as data warehouses, databases, and object stores, regardless of data structure.
“The platform is designed to allow teams to collaboratively develop models, where each model is represented as a configuration that can have multiple versions. Model differences and performance can be analyzed,” says Molino. explains.
Once the model meets the required performance criteria, it can be deployed as a REST endpoint for real-time predictions or for batch predictions using SQL-like queries with prediction capabilities.
The Importance of Low-Code in Machine Learning
The conversation then shifted to the importance of low-code development in implementing machine learning. Molino emphasized that simplifying the process is critical to broader industry adoption and improved return on investment.
By reducing development time from months to days, Predibase lowers the barrier to entry for organizations to experiment with new use cases and unlock potentially significant value.
“Organizations lose the incentive to explore worthwhile use cases when every project takes months or even years to develop. It’s very important,” Molino says.
“Using a low-code approach cuts development time to days and makes it easier to try different ideas and determine their value.”
LLM trends
The discussion also touched on the growing interest in large-scale language models. Molino acknowledged the great power of these models and their ability to change the way people think about AI and machine learning.
“These models are powerful and will revolutionize the way people think about AI and machine learning. In the past, data had to be collected and labeled before training a machine learning model. With APIs, we can directly query the model to get the predictions, which opens up new possibilities,” Molino explains.
However, Molino highlighted some limitations, including the cost and scalability of a per-query pricing model, relatively slow inference speeds, and data privacy concerns when using third-party APIs.
In response to these challenges, Predibase has introduced a mechanism that allows customers to deploy their models into a virtual private cloud, ensuring data privacy and giving them more control over the deployment process.
common mistakes
With more companies tackling machine learning for the first time, Molino shared his insights on common mistakes companies make. He emphasized the importance of understanding the data, use cases and business context before committing to development first.
“One common mistake is to have unrealistic expectations and a mismatch between what is expected and what is actually achievable. From a perspective, we are embarking on machine learning without a full understanding of the data and use cases,” Molino says.
Predibase addresses this challenge by facilitating hypothesis testing and providing a platform that integrates data understanding and model training to validate model suitability for specific tasks. Guardrails are in place so even inexperienced users can tackle machine learning with confidence.
The general availability of Predibase’s platform marks an important milestone in the company’s mission to democratize machine learning. Predibase aims to unlock the full potential of machine learning for both organizations and developers by simplifying the development process.
You can read the full interview with Molino below.
Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expos in Amsterdam, California and London. This event will be held concurrently with Digital Transformation Week.
