Generate Embeddings with AI Gateway

Generate text embeddings from a Databricks AI Gateway endpoint using the Databricks SDK.

When done, you will have:

• A configured embedding endpoint in your app environment
• A reusable helper function that generates vector embeddings from text input
• Integration with your existing Databricks workspace client via AppKit

Prerequisites

Verify these Databricks workspace features are enabled before starting. If any check fails, ask your workspace admin to enable the feature.

• Databricks CLI authenticated. Run databricks auth profiles and confirm at least one profile shows Valid: YES. If none do, authenticate with databricks auth login --host <workspace-url> --profile <PROFILE>.
• An embedding endpoint in Model Serving. Run databricks serving-endpoints list --profile <PROFILE> and confirm at least one embedding endpoint is listed (for example databricks-gte-large-en or databricks-bge-large-en, both 1024-dimension). Endpoint availability varies by workspace; note the endpoint name you plan to set as DATABRICKS_EMBEDDING_ENDPOINT.

Generate Embeddings with AI Gateway

Generate text embeddings from a Databricks AI Gateway endpoint using the Databricks SDK.

1. Find an embedding endpoint

databricks serving-endpoints list --profile <PROFILE>

Common embedding endpoints: databricks-gte-large-en (1024d), databricks-bge-large-en (1024d).

2. Configure environment

.env:

DATABRICKS_EMBEDDING_ENDPOINT=databricks-gte-large-en

app.yaml:

env:
  - name: DATABRICKS_EMBEDDING_ENDPOINT
    value: "databricks-gte-large-en"

3. Embedding helper

Create server/lib/embeddings.ts:

import { getWorkspaceClient } from "@databricks/appkit";

const workspaceClient = getWorkspaceClient({});

export async function generateEmbedding(text: string): Promise<number[]> {
  const endpoint =
    process.env.DATABRICKS_EMBEDDING_ENDPOINT || "databricks-gte-large-en";
  const result = await workspaceClient.servingEndpoints.query({
    name: endpoint,
    input: text,
  });
  return result.data![0].embedding!;
}

No additional dependencies — uses @databricks/appkit already in your project.

4. Verify

databricks serving-endpoints query <embedding-endpoint> \
  --json '{"input": "Hello, world!"}' \
  --profile <PROFILE>

Response includes a data array with embedding (float array).

References

• Query embedding models