​
Introduction

Eventual is a platform for processing multimodal data that gives you the serverless experience of a cloud data warehouse but for all the other modalities of data. Built on top of daft (our multimodal data engine), Eventual provides a unified API for processing petabytes of images, video, audio, and text without requiring distributed systems expertise.
Trusted by leading AI companies - Petabytes processed data daily at companies like Amazon, CloudKitchens, and Together AI

​
What is ev?

ev is the Python client library and command-line interface for the Eventual platform. It provides three core abstractions that make distributed multimodal data processing simple:

Jobs

Procedures composed of daft operations that scale automatically

Environments

Runtime configurations including dependencies and variables

​
Key Benefits

Write Python code that scales automatically. No need to learn complex distributed systems concepts or manage infrastructure.
Built-in handling for failures, retries, and edge cases. Your jobs run reliably even with network issues, API rate limits, and hardware failures.
10x faster than traditional engines for multimodal workloads. Optimized for processing images, videos, audio, and text at scale.
Seamlessly works with daft for data processing. Built by the team that created daft, ensuring perfect integration.

​
How It Works

# Simple job that processes millions of images
from ev import Env, Job

env = Env("3.11").pip_install(["daft==0.5.9"])
job = Job("process-images", env)

@job.main()
def process_images(bucket_path: str):
    import daft

    # Load images from S3 - automatically distributed
    df = daft.read_parquet(f"s3://{bucket_path}/images/*.parquet")

    # Apply ML model - scales across compute cluster
    df = df.with_column(
        "features",
        extract_features(df["image_data"])
    )

    # Save results - fault-tolerant writes
    df.write_parquet(f"s3://{bucket_path}/features/")
This simple Python script automatically handles distributed execution, fault tolerance, and scaling - no distributed systems expertise required!

​
Why Teams Choose Eventual

Built by engineers who’ve faced these problems firsthand. Our team comes from Databricks, AWS, Nvidia, Pinecone, GitHub Copilot - we’ve built the infrastructure powering self-driving cars, ML platforms, and AI applications at scale. We created Eventual because we were frustrated watching brilliant AI engineers waste months building distributed systems instead of solving actual problems. Now, companies processing petabytes daily trust us to handle their most critical workloads:
  • Amazon Retail - 40,000 years of compute time saved annually
  • Together AI - 10x speedup on 100TB+ text pipelines
  • MobilEye - 500x memory improvements for model training
Enterprise-Ready - Eventual runs in your cloud environment with minimal infrastructure setup, ensuring security and compliance while providing the scale you need.

​
Next Steps

Ready to get started? Here’s what to do next:

Quick Start

Get up and running in 5 minutes with our quick start guide

Installation

Install the ev SDK and configure your environment

Examples

See real-world examples of multimodal data processing

Core Concepts

Learn about Jobs, Environments, and Resources