How Lyft’s Data Issue Sparked the Creation of Eventual
▼ Summary
– Eventual founders identified a data infrastructure gap while working at Lyft’s autonomous vehicle program, where engineers struggled to process unstructured multimodal data efficiently.
– They developed Daft, a Python-native open source data processing engine designed to handle diverse data types (text, audio, video) quickly and reliably.
– Eventual was founded in early 2022, launched Daft’s open source version that year, and plans to release an enterprise product in Q3 2024.
– The company raised $27.5M in funding (seed and Series A) to expand its open source offering and build commercial AI applications for industries like robotics and healthcare.
– The multimodal AI market is projected to grow at 35% CAGR (2023–2028), with Daft positioned as a key solution for processing unstructured data in generative AI applications.
The challenges of managing unstructured data in autonomous vehicles led two engineers to develop a groundbreaking solution that now powers AI applications across industries.
During their time at Lyft’s autonomous vehicle division, Sammy Sidhu and Jay Chia encountered a critical bottleneck: the lack of a unified system to process the massive volumes of unstructured data generated by self-driving cars. From 3D scans and high-resolution images to audio logs and text inputs, engineers struggled to integrate disparate tools, wasting valuable time on infrastructure rather than innovation.
“We had brilliant minds working on autonomous vehicles, but they spent 80% of their time wrestling with data infrastructure,” Sidhu, now CEO of Eventual, recalled. The inefficiency sparked an idea, what if they built a tool to streamline multimodal data processing? After developing an internal solution for Lyft, Sidhu noticed widespread demand during job interviews, confirming the need for a scalable platform.
In 2022, Eventual launched Daft, an open-source Python engine designed to handle diverse data types, text, audio, video, and more, with speed and precision. Sidhu likens its potential impact to SQL’s revolution in tabular data, but for the unstructured data fueling modern AI. The timing proved prescient: ChatGPT’s explosion later that year highlighted the urgency for tools capable of managing multimodal inputs.
Industries beyond autonomous vehicles quickly adopted Daft, including robotics, healthcare, and retail tech. Early adopters like Amazon and Together AI validated its versatility. Eventual’s rapid traction attracted significant funding, $7.5 million in seed capital led by CRV, followed by a $20 million Series A round from Felicis, with backing from Microsoft’s M12 and Citi.
Astasia Myers of Felicis emphasized Eventual’s unique position in a market poised for explosive growth. “Annual data generation has surged 1,000x in two decades, with 90% being unstructured,” she noted. As generative AI expands into text, images, and voice, tools like Daft become indispensable.
With plans to enhance its open-source platform and launch a commercial product, Eventual aims to cement its role as the backbone of multimodal AI development, proving that sometimes, the biggest innovations arise from the most frustrating challenges.
(Source: TechCrunch)
One Comment