How a data processing problem at Lyft became the basis for Eventual

In the burgeoning landscape of artificial intelligence, a fundamental challenge often emerges: the efficient processing of vast, diverse datasets. This exact dilemma spurred the creation of Eventual, a company now at the forefront of multimodal data infrastructure, born from a critical need identified within Lyft’s autonomous vehicle program.

Eventual founders Sammy Sidhu and Jay Chia, both former software engineers at Lyft, observed firsthand the burgeoning data infrastructure problem. Self-driving cars, a quintessential example of complex AI applications, generate immense volumes of unstructured data, ranging from 3D scans and high-resolution photos to intricate text logs and audio recordings. At the time, Lyft engineers lacked a unified tool capable of understanding and processing these disparate data types simultaneously. This deficit often forced teams to painstakingly assemble various open-source solutions, a process fraught with inefficiency and reliability concerns.

“We had all these brilliant PhDs, brilliant folks across the industry, working on autonomous vehicles but they’re spending like 80% of their time working on infrastructure rather than building their core application,” remarked Sidhu, now Eventual’s CEO, in a recent interview. He emphasized that the majority of these bottlenecks revolved around data infrastructure.

Recognizing this critical gap, Sidhu and Chia collaborated to develop an internal multimodal data processing tool for Lyft. The transformative potential of this solution became strikingly clear to Sidhu when, during subsequent job interviews, potential employers consistently inquired about developing similar data solutions for their own organizations. This widespread interest validated the universal nature of the problem and laid the groundwork for Eventual’s inception.

Eventual’s flagship product is Daft, a Python-native, open-source data processing engine meticulously engineered for rapid performance across diverse modalities including text, audio, video, and more. Sidhu articulates an ambitious vision for Daft: to revolutionize unstructured data infrastructure in the same profound way SQL transformed tabular datasets.

Founded in early 2022, nearly a year prior to the mainstream explosion of ChatGPT, Eventual demonstrated remarkable foresight regarding the impending demand for robust data infrastructure. The initial open-source version of Daft was launched in 2022, with an enterprise product slated for release in the third quarter of the current year. The surge in AI applications post-ChatGPT, particularly those integrating images, documents, and videos, dramatically amplified Daft’s usage and relevance.

While Daft’s origins are rooted in the autonomous vehicle sector, its utility extends across numerous industries grappling with multimodal data, including robotics, retail technology, and healthcare. Eventual currently boasts an impressive roster of customers, including industry giants like Amazon, CloudKitchens, and Together AI.

The company has recently secured significant investment, completing two funding rounds within an eight-month period. An initial $7.5 million seed round was led by CRV, followed by a substantial $20 million Series A round spearheaded by Felicis, with key participation from Microsoft’s M12 and Citi. This latest capital injection is earmarked for bolstering Eventual’s open-source offerings and developing its commercial product, which will empower customers to build advanced AI applications atop their processed data.

Astasia Myers, a general partner at Felicis, disclosed that her firm discovered Eventual through a comprehensive market mapping exercise focused on identifying data infrastructure solutions capable of supporting the escalating number of multimodal AI models. Myers highlighted Eventual’s distinction as a first-mover in this rapidly evolving domain, citing the founders’ direct experience with the data processing problem as a significant differentiator. She underscored that Eventual is actively addressing a rapidly expanding market need.

Indeed, the multimodal AI industry is projected for exponential growth, with a predicted 35% compound annual growth rate (CAGR) between 2023 and 2028, according to management consulting firm MarketsandMarkets. This growth is underpinned by an unprecedented surge in data generation: annual data generation has increased 1,000-fold over the past two decades, with an astounding 90% of the world’s data produced within the last two years, and, according to IDC, the vast majority of this data remains unstructured. Myers concludes that “Daft fits into this huge macro trend of generative AI being built around text, image, video, and voice. You need a multimodal-native data processing engine.”