Baseten | Self-Serve Apps for ML Teams

TL;DR

Base 10's platform reduces the time to value of machine learning efforts by allowing teams to build scalable user-facing workflow integrated apps powered by AI.

Transcript

hi everyone welcome back to gray matter i'm sarah gua a general partner at greylock it remains very difficult to build end-to-end products that leverage machine learning for business impact and base 10 makes it fast seamless and serverless to build scalable user-facing workflow integrated apps powered by ai we've been working with the base 10 team ... Read More

Key Insights

• 🎰 Base 10's platform aims to reduce the barrier for companies to leverage machine learning and derive value from their models.
• 🧡 The platform offers tools for hosting, business logic integration, and UI building, catering to a wide range of machine learning application development needs.
• 🎰 By targeting data science and machine learning teams, Base 10 aims to democratize machine learning and empower more companies to benefit from its applications.
• 🤪 Base 10's focus on going from model to business value fills a gap in the market, where research and innovation in machine learning often outpace the practical application of those advancements.
• 👻 The platform's modular format allows users to choose and integrate specific components according to their needs, providing flexibility in leveraging the platform's functionalities.
• 🔶 Base 10's platform has been adopted by various companies, ranging from startups to larger enterprises, for diverse use cases, including content moderation, underwriting, and data labeling.
• 📌 The company operates as a remote-first organization, leveraging the advantages of hiring talent from diverse geographic locations.
• 😒 Base 10 aims to build an ecosystem around its platform, empowering developers and data scientists to collaborate and create templates that address specific use cases.

Questions & Answers

Q: What does the Base 10 platform offer in terms of machine learning application development?

The Base 10 platform simplifies the process of incorporating machine learning models into production-grade applications, reducing the time to value from months to hours. It provides tools for hosting models, integrating business logic, and building UI for the applications.

Q: How does Base 10 differentiate itself from other machine learning tools in the market?

While there are many startups and existing tools in the machine learning space, Base 10 focuses on going from model to business value, rather than just hosting models behind an API. They aim to make machine learning more of a product function rather than a research function, unlocking the potential for real-world business applications.

Q: What prompted Base 10 to build their platform in such a competitive landscape?

Base 10 saw an opportunity in the market to bridge the gap between machine learning research and real business applications. They recognized that many companies were early in their journey of applying machine learning to create value, and there was a lack of focus on this aspect in the industry. They wanted to create a tool that would simplify and streamline the process of building machine learning applications.

Q: Who is the target audience for the Base 10 platform?

Base 10 targets data science and machine learning teams that have the expertise to create models but lack the resources and support to fully integrate them into production-grade applications. They aim to empower these teams to build applications that add value to their businesses.

Summary

In this video, Sarah Guo, a general partner at Greylock, interviews Tuhin Sinha, the co-founder and CEO of Base 10. Base 10 is a platform that aims to simplify the process of building and deploying machine learning applications. Tuhin discusses how Base 10 reduces the time to value of machine learning efforts by enabling teams to incorporate their machine learning models into production-grade applications within hours instead of months. He also talks about the challenges of building machine learning applications and the white space in the industry that Base 10 aims to fill. Tuhin shares his background and experience in the field of machine learning and explains why it is important to democratize access to machine learning tools. He talks about the target audience for Base 10 and the range of use cases it can support. Tuhin also discusses the company's journey, from early adopters to the current beta release, and the challenges and opportunities of building a remote team. He shares the vision for Base 10's future, including building an ecosystem of templates and increasing collaboration and effectiveness in the machine learning process.

Questions & Answers

Q: Can you explain what the Base 10 platform does?

Base 10 reduces the time to value of machine learning efforts and enables teams to incorporate their machine learning models into production-grade applications within hours. It provides a single end-to-end platform where data science and ML teams can preprocess data, serve their models, and build user interfaces to create full machine learning applications.

Q: Where does Base 10 sit in the landscape of machine learning tools?

Base 10 focuses on going from model to business value, rather than just providing tools for model deployment. It aims to simplify the process of building and deploying machine learning applications by providing a more integrated and streamlined platform. While there are many existing tools in the machine learning space, Base 10 saw an opportunity to fill the white space in the industry and provide a tool that enables data scientists and ML teams to create value-added applications.

Q: What convinced you that there was something important still to be built in this space?

Base 10 saw an opportunity to fill a gap in the machine learning landscape by focusing on applying machine learning to real business problems. While many companies have made significant contributions to model architecture and techniques, there is still a lack of focus on applying those innovations to real-world problems. Most companies are still in the early stages of applying machine learning and have not fully realized the potential value. Base 10 aims to provide the tooling and platform to unlock that value and bridge the gap between research and application.

Q: Can you talk about your background and the experiences that led to the development of Base 10?

Tuhin, along with his co-founders, Bill and Amir, have backgrounds in data science and machine learning. They have worked on various projects where they saw the challenges of integrating machine learning models into real-world applications. Tuhin shared his experience at Gumroad, where they faced fraud issues and had to build a machine learning model to solve the problem. However, they realized that building the model was only one part of the equation. They had to learn full-stack engineering to connect the model to business value. This experience, along with similar challenges at other companies, led them to start Base 10 and create a platform that simplifies the process of building machine learning applications.

Q: Who is the target audience for Base 10 and what kind of use cases can it support?

Base 10 targets data science and machine learning teams that have the expertise to create machine learning models but lack the resources to build the necessary infrastructure for deployment and integration. The platform is designed to enable these teams to build applications that add value using machine learning. It supports a wide range of use cases, from content moderation and data labeling to user verification and energy grid placement. Base 10 aims to empower data scientists and ML teams to build applications regardless of their company's size or resources.

Q: How does an organization adopt Base 10 and who are some of the early adopters?

Organizations typically start with one data scientist or engineer trying out Base 10 and building an application. Once they see the value, they bring in other team members or users to collaborate on the platform. The adoption process varies depending on the use case, but the goal is to empower individuals to build applications and iterate on them. Some early adopters of Base 10 include companies like Patreon and Pipe, who use it for content moderation and underwriting assets. There are also non-profits using it for translating coded materials and placing energy grids offshore.

Q: How has the company grown and what is the current state of Base 10?

Base 10 has grown to a team of around 20 people, with a small office in San Francisco but primarily working remotely. The company initially wasn't planning to be remote, but the pandemic allowed them to hire talent from different locations. They have been able to build a remote culture and collaborate effectively. The platform is currently in the beta release phase, as the team continues to work on improving usability and building out the different components of the platform. The goal is to reach parity across all three pillars of the platform—model deployment, workflow building, and UI building—before a general availability release.

Q: Why is Base 10 released as a beta and what are the future plans for extending the platform?

Base 10 is released as a beta to invite users in and set expectations for continued improvement and development. The team wants to surprise users with the value they can achieve on the platform. The beta release also allows the team to refine and perfect the abstractions and features of the platform. The three pillars of Base 10—the model deployment engine, workflow builder, and UI builder—provide opportunities to build out an ecosystem and platform. The goal is to create an ecosystem of templates that combine models, workflows, and UI components to streamline the development of specific use cases.

Q: What change do you hope to see in the world if Base 10 succeeds?

Tuhin hopes that successful adoption of Base 10 will reduce the costs of shipping ML and increase collaboration and iteration cycles. He believes this will lead to more models being used by more businesses and an overall increase in the effectiveness of machine learning efforts. By simplifying the process of building machine learning applications and making them more accessible, Base 10 aims to enable organizations to leverage the power of ML and drive real-world impact.

Summary & Key Takeaways

• Base 10's platform allows teams to incorporate machine learning models into production-grade applications within hours instead of months.

• The platform simplifies the process of hosting, integrating business logic, and building UI for machine learning applications.

• Base 10 is targeting data science and machine learning teams that lack the resources to build and deploy their models effectively.