Sammy Sidhu and Jay Chia first encountered a significant data infrastructure challenge while working as software engineers on Lyft’s autonomous vehicle program. The pair observed firsthand that self-driving cars generate immense quantities of unstructured data — 3D scans, photographs, text, and audio — but no single tool then available could adequately handle all these diverse types of data simultaneously and cohesively. As a result, Lyft’s engineers were forced to stitch together disparate open-source tools, an inefficient method prone to reliability concerns.
Recognizing this obstacle, Sidhu and Chia built an internal multimodal data-processing solution at Lyft. This experience laid the groundwork for their startup, Eventual. Sidhu noted in a recent conversation that highly talented engineers and PhDs were spending the vast majority of their time addressing underlying data infrastructure issues rather than developing their core software applications. He realized the issue extended far beyond Lyft when other companies began expressing interest during his job search, repeatedly inquiring about similar data-processing systems.
Officially founded in early 2022, Eventual created Daft, an open-source, Python-native data-processing engine designed explicitly to handle multimodal unstructured data from text and audio to images and video. Sidhu aims to position Daft as a revolutionary shift in managing unstructured data, analogous to how SQL transformed the handling of structured tabular datasets in previous decades.
Daft launched its initial open-source version in late 2022, almost a full year ahead of the mainstream generative AI boom triggered by ChatGPT’s release. The founders soon found their timing ideal: the arrival of ChatGPT and subsequent widespread adoption of generative AI boosted the demand for multimodal AI solutions significantly. Sidhu explained that once developers began actively exploring AI applications involving images, documents, and videos, Daft saw a remarkable rise in adoption.
Though initially inspired by autonomous vehicle technology, Eventual’s data processing engine has quickly gained traction across multiple sectors that rely heavily on multimodal datasets, including healthcare, retail technology, and robotics. Major customers of Eventual now include Amazon, CloudKitchens, and Together AI, among others.
Investors quickly recognized the value of Eventual’s innovative approach. In rapid succession, Eventual secured two funding rounds within eight months: a $7.5 million seed round led by CRV and a subsequent $20 million Series A round led by Felicis, with participation from Microsoft’s M12 and Citi. The investments will support further development of Daft’s open-source offerings and finance a new enterprise product set to launch in the third quarter, designed to help businesses efficiently create multimodal AI applications based on processed data.
Astasia Myers, general partner at Felicis, initially identified Eventual during a strategic market assessment of companies poised to capitalize on the growing multimodal AI sector. Myers pointed to Eventual’s early market entry and the founders’ hands-on familiarity with data infrastructure problems as key reasons for investment. The broader market context underscores her enthusiasm: according to forecasts by consulting firm MarketsandMarkets, the multimodal AI market is expected to experience compound annual growth of around 35% between 2023 and 2028.
In highlighting the potential of multimodal data solutions, Myers cited significant global trends in data production: data generation has increased a thousand-fold over the past twenty years, and an estimated 90% of current global data has been created within just the last two years. Unstructured data, Myers emphasized, dominates this rapid sea of information, underscoring the critical need for solutions like Daft.
With their technology timed perfectly to capitalize on the surging demand for multimodal data processing, Eventual is uniquely positioned in an increasingly competitive landscape to play a leading role in reshaping data infrastructure across diverse industries.
