April 2022 Summaries

At Baseten, the company co-founder started to solve issues faced by data science teams building machine learning models in companies of all sizes. They realized that despite everyone being on board with machine learning, most teams were ill-equipped due to skills, resources, and credibility issues. Baseten aimed to build a general-purpose toolkit to leverage data scientists' skills and turn them into full-stack shippers by providing modular pillars for deploying models, setting up backend infrastructure, and building UIs. The company quietly announced its public beta after 18 months of building, with the help of early users, and is now launching it to allow teams to build full-stack applications without needing to become engineers. Baseten offers a set of building blocks for each step of the journey, including model deployment, integration, design, and shipping, which can lower the cost of using machine learning. The company has raised over $20 million in funding led by Sarah Guo from Greylock Partners and is actively hiring for roles across engineering, design, and go-to-market to amplify and accelerate the impact of machine learning.

When you deploy a model on Baseten, you can call it via an API endpoint with zero configuration, saving the effort of building and deploying a model server. This allows for faster interaction with models, but may require additional glue code to integrate generative AI models with applications. The API endpoint is available behind every model in your Baseten workspace, accessible through the "Call model" button on the model dashboard or by calling it from the command line, programming language of choice, or an API tool like Postman. The API endpoint can be used to parse model output and use it in production, with instructions provided for integration into various applications, including Python, cURL, JavaScript, and Truss.