Advanced Deployment Options

Slai

Slai is the fastest way for developers to build machine learning applications that can be deployed on serverless infrastructure and shared with anyone in the world.

What is Slai?

Train

Handler

Interactively testing a model in the sandbox.

Testing

Datasets

An introduction to model revisions on Slai.

Revisions

How to add dependencies to your environment

Dependencies

Your model can be deployed as a real-time API. Slai includes client libraries for cURL, Python, and Node, allowing you to call the deployed model from your own codebase. 

Deployments

Import Code from Github

Shipping metrics to a third party tool is fairly straightforward. In this example, we'll log our training metrics to Weights and Biases (WandB) to track our experiments.

Shipping Metrics Outside Slai

Slack Notifications

Slai comes with several integrations to import data from remote data stores, like S3 buckets. This guide explain how to connect an S3 bucket and use it in your sandbox.

Connecting S3 Buckets

Slai supports direct integration with Google BigQuery. This guide demonstrates how to retrieve data from BigQuery for use in a Slai Sandbox.

Connecting to Google BigQuery

Already have a model you'd like to migrate to Slai? Here's how to port it over.

Migrating an existing model

Here's a quick guide on how to use pre-trained models in Slai.

Using Pre-Trained Models

Custom Docker Images

Recommendation System

Slai charges usage based billing for the compute used to train and deploy your apps.

Tracking Usage

API Keys

The model class is the entry point for your deployed model.

Model

Usage-Based Billing

Debugging Guide

Privacy Policy

Beam is a cloud platform where you can provision infrastructure, develop on remote runtimes from your local machine, and deploy apps as serverless functions — without leaving your IDE

Introduction

Get started with Beam in just a few minutes

Quickstart

Install Beam

A short tutorial on building a sentiment analysis model and deploying it as a REST API with Beam

Deploying an ML Model

Let's build a simple web scraper which extracts headlines from The New York Times and uses a BERT model from Huggingface to detect the sentiment of each.

Running a Web Scraper

Building an API to generate images with stable diffusion

Stable Diffusion API

You can integrate Beam into an existing CI/CD process to deploy your code automatically.

Integration into CI/CD

REST API

Webhook

Scheduled Cron Jobs

Read and write files from inside a Beam environment

Outputs

A readable and writable data store that will persist across tasks

Persistent Volumes

After deploying, you can view the traffic and container logs on your task in the dashboard.

Logging and Monitoring

Build and deploy Beam apps directly from the command line

The library for creating and managing Beam apps

API keys are used to access the CLI and SDK

Beam includes a secrets manager that allows you to use secrets across your projects.

Home

Sandbox

Integrations

User Guides

Dashboard

API Reference

Pricing

Resources

Security

Advanced Deployment Options