How the problem of data processing in Lyft has become the basis for the basis

when ultimate The founders of Sammy Sidhu and Jay Chia were working as a software engineer in Lyft’s Mostronmous vehicles, they have seen a problem with fermentation data – which will become larger with the rise of artificial intelligence.

Self -driving cars produce a lot of non -structured data from 3D scanning and sound to the text and sound. There was no tool for Lyft engineers to that could understand and process all of these different types of data at the same time – all in one place. Leave this engineers to collect open source tools in a long process with reliability problems.

“We had all this great doctorate, the wonderful people throughout the industry, and work on independent cars, but they spend like 80 % of their time working on infrastructure rather than building their basic application,” said Sidho, CEO of Techcrunch in a recent interview. “Most of these problems they face were about data infrastructure.”

Help Sidhu and Chia build Lyft multimedia data processing tool. When Sidhu started applying for other jobs, he found the interviews who continued to ask about the possibility of building the same data solution to their companies, and the idea was born behind the final.

It eventually built Python open -source data processing engine, known as DAFT, designed to work quickly through different exhibitions of text to sound and video, and more. Sidhu said that the goal is to make DAFT as an infrastructure for non -structured data such as SQL was for tabular data groups in the past.

The company was founded in early 2022, nearly a year before the Chatgpt version, and before many people realize this data infrastructure gap. They launched the first open source version of DAFT in 2022 and are preparing to launch the Enterprise product in the third quarter.

“The Chatgpt explosion, what we have seen is just many other people who then build artificial intelligence applications with different types of methods,” said Sedho. “Then everyone started using things like photos, documents and videos in their applications. This is a type of place we saw, increased use significantly.”

While the original idea behind construction stems from the area of ​​self -government vehicles, there are many other industries that treat multimedia data, including robots, retail and health care technology. The company is now Amazon, Cloudkitchens and Tower AI, among other things, as customers.

I recently raised two rounds of financing within eight months. The first was the seed tour of $ 7.5 million led by CRV. Recently, the company raised a $ 20 million tour led by Felisis with the participation of M12 and Citi from Microsoft.

This last round will go to the ultimate open source width in addition to creating a commercial product that allows its customers to build artificial intelligence applications from these processing data.

Istizia Maires, the general partner of Velicis, TECHRUNCH to eventually find it through the mapping exercise in the market, which included the search for infrastructure for data that will be able to support the increasing number of multimedia artificial models.

Myers said that ultimately has emerged because it was the first engine in space – which is likely to become more crowded – and based on the fact that the founders have dealt with this data processing problem directly. She added that ultimately solves a growing problem.

The multimedia intelligence industry is expected to grow in A. 35 % compound annual growth rate Between 2023 and 2028, according to the MarketsandMarkets Administrative Consulting Company.

“The annual data generation has increased 1000X over the past twenty years and 90 % of the world’s data has been created in the past two years, and according to IDC, the vast majority of data is not organized,” Mayers said. “DAFT is suitable for this huge total trend of the tweede that is built around text, image, video and sound. You need a multimedia data processing engine.”