# Eventual Scales Multimodal Engine Daft on $30M to Replace Big Data Stacks in AI

**Author:** Zero G Talent
**Published:** 2026-11-13
**Category:** engineering
**Tags:** yc-hiring, data infrastructure, multimodal ai, distributed systems, rust, software engineering

[Read on Zero G Talent](https://zerogtalent.com/blog/eventual-scales-multimodal-engine-daft-on-30m-to-replace-big-data-stacks-in-ai)

---

Modern foundation models, perception pipelines, and autonomous vehicle stacks consume petabytes of raw images, video, audio, and sensor streams. Traditional data platforms like Spark, Snowflake, and Databricks were architected for tabular numbers and text records, leaving machine learning teams to assemble fragile pipelines that fail when handling complex data types.

[Eventual](https://zerogtalent.com/ai-companies/eventual) is building an engine to fix that foundation. The San Francisco company develops Daft, an open-source distributed data engine written in Rust and exposed through a Python API. Daft makes working with multimodal data—including video feeds, lidar scans, and high-frequency telemetry—as direct as executing relational queries, while streaming directly into GPU memory for training and inference. Backed by $30 million in venture funding, including a $20 million Series A led by Felicis, Eventual is doubling its team in San Francisco's Mission District to scale both its open-source project and its enterprise cloud platform.

## Rethinking Data Engines for Video, Audio, and Tensors

When data systems process structured enterprise records, aggregations cause data to shrink over time: hundreds of gigabytes of logs condense into a compact dashboard table. Multimodal AI workloads do the opposite. A single terabyte of pointers and URLs inflates by orders of magnitude when downloaded, decoded into uncompressed image frames, and converted into high-dimensional tensors.

Legacy engines built on Java Virtual Machines frequently crash under this memory expansion. When Eventual co-founders [Sammy Sidhu](https://zerogtalent.com/people/sammy-sidhu) and [Jay Chia](https://zerogtalent.com/people/jay-chia) worked as software engineers on Lyft's Level 5 autonomous driving program, they watched machine learning engineers spend up to 80% of their time troubleshooting distributed infrastructure rather than refining models. As Sidhu told [TechCrunch](https://techcrunch.com/2025/06/24/how-a-data-processing-problem-at-lyft-became-the-basis-for-eventual/), autonomous vehicles generate 3D scans, video feeds, and audio that existing tools could not process in one unified framework.

Sidhu and Chia designed Daft around dynamic streaming execution. Rather than pulling massive static partitions into RAM, Daft inspects pipeline operators and dynamically sizes batches, pipelining downloads, decoding, and model inference. The execution engine is written in Rust for low-level memory control and binds directly to distributed runtimes like Ray, storage formats like Apache Iceberg and Delta Lake, and cloud object stores like AWS S3.

In benchmarks presented at [The AI Conference™](https://www.youtube.com/watch?v=NMan0VachXI), Chia demonstrated Daft embedding one million images stored in AWS S3 in three minutes on commodity Nvidia hardware, classifying the same million images across vision-language models in four minutes, and chunking and embedding one million PDF documents in 12 minutes.

Today, Daft handles production workloads at organizations including Amazon, Together AI, Mobileye, and CloudKitchens. Alongside the open-source engine, Eventual is rolling out Eventual Cloud, a managed platform designed to orchestrate multimodal indexing, pipeline failures, and line-rate GPU feeding without infrastructure overhead.

## Autonomous Driving and Systems Heritage

Eventual’s leadership brings deep experience in high-performance computing, distributed storage, and machine learning operations.

CEO [Sammy Sidhu](https://zerogtalent.com/people/sammy-sidhu) studied Electrical Engineering and Computer Sciences at UC Berkeley before conducting medical AI research and high-frequency trading. He later served as Chief Architect at DeepScale, an embedded computer vision startup acquired by Tesla, before joining Lyft Level 5 to work on distributed training infrastructure, deep learning compilers, and distributed data loaders.

Co-founder [Jay Chia](https://zerogtalent.com/people/jay-chia) graduated from Cornell University with a degree in Computer Science, conducting research in computational biology and deep learning. Prior to co-founding Eventual, he built large-scale machine learning and data platforms across biotechnology at Freenome and autonomous driving systems at Lyft Level 5.

After graduating from Y Combinator's W22 batch, the team raised a $10 million seed round led by CRV, with participation from Array Ventures and Essence VC, followed by a $20 million Series A led by Felicis, with participation from Microsoft's M12 and Citi, as reported by [SiliconANGLE](https://siliconangle.com/2025/06/24/eventual-launches-30m-funding-streamline-multimodal-data-processing/). Additional angel investors include co-founders of Databricks and Perplexity.

## Engineering Roles in the Mission District

Eventual has grown to 18 employees, drawing talent from Databricks, AWS, Nvidia, Pinecone, GitHub Copilot, and Tesla. The company is now actively hiring across its core engineering teams to support enterprise deployments and core engine development. All roles operate on an onsite schedule of four days per week in Eventual's San Francisco office in the Mission District.

On the core infrastructure side, the company is hiring a [Software Engineer, Data Systems](https://zerogtalent.com/ai-jobs/eventual/software-engineer-systems-79973) ($150,000 - $250,000). The role centers on creating distributed storage and a high-throughput data engine capable of handling petabytes of video, lidar, and high-frequency telemetry data. Engineers work on real-time perception indexing, distributed compute primitives, and loading data at line-rate into GPUs for training and inference.

To build product layers and streaming architecture, Eventual lists a [Software Engineer, Multimodal Backend Systems](https://zerogtalent.com/ai-jobs/eventual/software-engineer-multimodal-backend-systems-114722) ($150,000 - $250,000). Responsibilities include developing core architecture, real-time video infrastructure, physical AI and robotics telemetry handling, and multimedia customer frontends.

To advance model-driven processing within data pipelines, the team is recruiting a [Research Engineer, Multimodal Data](https://zerogtalent.com/ai-jobs/eventual/research-engineer-multimodal-data-94775) ($150,000 - $300,000) for its Visual Understanding group. This engineer owns the layer that makes petabytes of raw video searchable by visual content, selecting and training models that make corpus-scale annotation tractable and structuring datasets that train customer models.

## Open Positions at Eventual

| Role | Location | Work Mode | Compensation |
| --- | --- | --- | --- |
| [Software Engineer, Multimodal Backend Systems](https://zerogtalent.com/ai-jobs/eventual/software-engineer-multimodal-backend-systems-114722) | San Francisco, CA, US | Onsite (4 days/week) | $150,000 – $250,000 |
| [Research Engineer, Multimodal Data](https://zerogtalent.com/ai-jobs/eventual/research-engineer-multimodal-data-94775) | San Francisco | Onsite (4 days/week) | $150,000 – $300,000 |
| [Software Engineer, Data Systems](https://zerogtalent.com/ai-jobs/eventual/software-engineer-systems-79973) | San Francisco, CA, US | Onsite (4 days/week) | $150,000 – $250,000 |

[See all open roles at Eventual](https://zerogtalent.com/ai-companies/eventual)

---
*Source: [Zero G Talent](https://zerogtalent.com), the job board for space, defense, AI, robotics and frontier-tech companies.*