Worker

This document covers the worker and job execution system that manages agent processes, handles backend registration, and coordinates job assignment and execution. This system provides the foundation for running VideoSDK agents either locally or as part of a distributed backend infrastructure.

Architecture Overview

The worker and job system consists of three primary components that work together to execute agent code:

• WorkerJob: The main entry point that configures and starts agent execution
• Worker: Manages process pools, backend communication, and job lifecycle
• JobContext: Provides runtime context and resources to agent entrypoint functions

Core Components

Worker Class

The Worker class manages the complete lifecycle of agent execution, including process management, backend communication, and job coordination.

Core Responsibilities:

• Process pool management and lifecycle
• Backend registry communication
• Job assignment and execution coordination
• Resource monitoring and cleanup
• Error handling and recovery

WorkerJob

The WorkerJob class serves as the primary entry point for creating and running agents. It accepts an entrypoint function and configuration options, then delegates to the Worker class for execution.

from videosdk.agents import WorkerJob, Options, JobContext, RoomOptions

# Configure worker options
options = Options(
    agent_id="MyAgent",
    max_processes=5,
    register=True,   # Registers worker with backend for job scheduling
)

# Set up room configuration
room_options = RoomOptions(
    name="My Agent",
)

# Create job context
job_context = JobContext(room_options=room_options)

# Define your agent entrypoint
async def your_agent_function(ctx: JobContext):
    # Your agent logic here
    await ctx.connect()
    # Agent implementation...

# Create and start the worker job
job = WorkerJob(
    entrypoint=your_agent_function,
    jobctx=lambda: job_context,
    options=options,
)

job.start()

• Entrypoint: An async function that serves as your agent's main execution logic. This function receives a JobContext parameter and contains your agent implementation.

• JobContext: Provides the runtime environment for your agent, managing room connections and VideoSDK integration. It handles room setup, authentication, and cleanup operations.

• Options: Configuration settings for worker execution including process management, authentication, and backend registration. You can find worker options here ↗.

Key Methods:

• start(): Initiates worker execution based on configuration

Deployments

Choose how to deploy your VideoSDK agents based on your infrastructure needs and requirements.

Agent Cloud

Deploy your agents to VideoSDK's managed cloud infrastructure

Self Hosting with Worker

Learn how to deploy workers for self-hosted agent infrastructure

Examples - Try out yourself

We have examples to get you started. Go ahead, try out, talk to agent, understand and customize according to your needs.

Worker Example

Checkout worker implementation

AI Telephony Example

Worker usage in telephony