Hey followers! Let me tell you about an interesting tech story involving Lyft and some brilliant minds at Eventual.
While working on Lyft’s autonomous car program, Sammy Sidhu and Jay Chia saw a massive data infrastructure challenge. Self-driving cars generate heaps of unstructured data—think 3D scans, images, audio, and text—and no single tool could handle it all seamlessly. Engineers had to cobble together various open-source solutions, leading to unreliable and time-consuming workflows.
Sidhu, who is now Eventual’s CEO, shared that many talented PhDs wasted much time on infrastructure instead of core innovations. To solve this, they created an internal multimodal data processing tool at Lyft. Interestingly, Sidhu then found that many interviewers asked him if he could build a similar tool for their companies, sparking the idea for Eventual.
Eventual built Daft, a Python-based open-source engine capable of swiftly processing varied data types—from text and images to audio and video. Sidhu envisioned Daft would be as revolutionary for unstructured data as SQL was for spreadsheets. Founded in early 2022, the company launched Daft’s initial version before ChatGPT became popular, and is planning its enterprise launch soon.
The rise of generative AI applications using multimodal data—like videos, images, and speech—has boosted demand for tools like Daft. Though it started in autonomous vehicles, the technology now finds use in robotics, healthcare, and retail, with clients like Amazon and CloudKitchens. Eventual recently secured $27.5 million in funding to expand its open-source platform and develop commercial products, addressing the fast-growing multimodal AI industry forecasted to grow at 35% annually.
Felicis partner Astasia Myers highlighted how Eventual’s firsthand experience with data problems made it a standout in the evolving AI infrastructure landscape. With data generation soaring and most of it unstructured, Daft is positioned to meet the demands of the expanding multimodal AI ecosystem.