FAQs

General

1. What is the Sahara AI Product Suite?

Sahara AI is a decentralized platform designed to manage and deploy AI services. It provides robust tools for developers, data providers, and AI innovators, enabling them to build, manage, and monetize AI assets through a secure, scalable infrastructure that integrates blockchain, smart contracts, and decentralized execution.

2. What are the core components of the Sahara AI Product Suite?

The platform is composed of several integrated hubs:

• Developer Portal: Your gateway to access and manage all services.

• Data Hub: A marketplace for discovering, purchasing, and selling datasets.

• Model Hub: A centralized interface to create, browse, and deploy AI pipelines (supporting both Retrieval-Augmented Generation and prompt-based pipelines).

• Compute Hub: For managing compute instances and deploying AI models.

• Data Vaults: Secure, decentralized storage for datasets.

• Developer APIs: For programmatic interaction with the platform’s services.

Onboarding

1. How do I get started with Sahara AI?

Begin with the Instructions for Testing :

Step 1: Add data to your Vaults either from the Data Hub or by uploading personal data.

Step 2: Set up an endpoint in the Compute Hub.

Step 3: Create and deploy your AI pipeline via the Model Hub.

Step 4: Integrate your endpoint and monitor performance through the Metrics Dashboard.

You can follow these workflows using either the Graphical User Interface (GUI) or the Developer APIs.

2. How can I add data to my Vault?

You have two options...

From the Data Hub:

1. Browse available datasets.

2. Purchase a dataset.

3. Import the dataset into your selected Vault.

Upload Personal Data:

1. Navigate to the Vaults tab.

2. Create a new Vault by providing a name and description.

3. Upload your files (supported formats include CSV, JSON, and Parquet).

4. Configure metadata tags for better discoverability and provenance tracking.

What file formats are supported for uploading personal data?

Sahara AI supports CSV, JSON, and Parquet formats for personal data uploads, ensuring compatibility with a wide range of datasets.

3. How do I set up an endpoint in the Compute Hub?

Follow these steps:

1. Open the Compute Hub tab on the Sahara dashboard.

2. Click the New Endpoint button.

3. Fill out the configuration form by selecting a compute provider (e.g., Lepton, Predibase, Sagemaker, Bedrock, or OpenAI), choosing a model, and naming your instance.

4. Review your configuration and click Create Instance to launch the endpoint.

4. What is the process for creating and deploying pipelines in the Model Hub?

The process involves two main steps:

Creating a Pipeline

1. Navigate to the Model Hub and click Create Pipeline.

2. Choose between a Retrieval-Augmented Generation (RAG) Pipeline (which requires Vault integration) or a prompt-based pipeline.

3. Configure the pipeline by providing details such as name, description, instructions, and (if applicable) selecting Vaults for data retrieval.

Deploying the Pipeline

1. Publish the pipeline.

2. Select a compatible compute provider.

3. An API endpoint is automatically generated for integration and further testing.

Technical

1. What is the difference between RAG pipelines and prompt-based pipelines?

RAG Pipeline:

Integrates data retrieval from Vaults, making it ideal for tasks that require referencing external data.

Prompt-Based Pipeline:

Designed for conversational or structured AI tasks, allowing you to interact with the AI through specific instructions or prompts.

2. How do I test and monitor my deployed pipeline?

Testing the API:

Use the generated API endpoint to send sample requests (e.g., using a JSON payload like {"input": "What is the weather today?"}) and review the AI responses.

Monitoring Metrics:

Access the Metrics tab in your pipeline’s details page to monitor usage metrics such as token usage, API calls, and overall system performance.

3. What compute providers are supported by Sahara AI?

Sahara AI supports multiple compute providers including Lepton, Predibase, Sagemaker, Bedrock, and OpenAI. This variety allows you to choose the provider that best meets the requirements of your AI workload.

4. How can I interact with Sahara AI programmatically?

Developers can use the provided Developer APIs to:

• Create, deploy, and manage AI pipelines.

• Control compute resources.

• Integrate Sahara AI services directly into their applications, ensuring seamless automation and customization of workflows.

Additional Support

1. Where can I find additional resources and support?

Additional documentation, tutorials, and support can be accessed through the Developer Portal on the Sahara AI platform. For further assistance, reach out to the Sahara AI support team via the platform’s slack channel, which you've been invited to.