Question: Can you make a DynamoDB table case-insensitive?

Answer

No. DynamoDB has no collation or case-insensitivity setting. It compares strings by the bytes of their UTF-8 encoding, so key lookups, sort key conditions, begins_with, contains and filter expressions are all case-sensitive: Alice and alice are different partition key values.

The standard workaround is to store a normalized copy of the value and query the copy:

import { DynamoDBClient } from "@aws-sdk/client-dynamodb";
import { DynamoDBDocumentClient, GetCommand, PutCommand } from "@aws-sdk/lib-dynamodb";

const ddb = DynamoDBDocumentClient.from(new DynamoDBClient({}));
const normalize = (email) => email.trim().toLowerCase();

// Write: key on the normalized value, keep the original for display
const email = "Alice@Example.com";
await ddb.send(
  new PutCommand({
    TableName: "Users",
    Item: { pk: `USER#${normalize(email)}`, email },
    ConditionExpression: "attribute_not_exists(pk)",
  })
);

// Read: normalize the search term the same way
const { Item } = await ddb.send(
  new GetCommand({ TableName: "Users", Key: { pk: `USER#${normalize("alice@example.COM")}` } })
);

If the value isn't part of the primary key, store the lowercased copy in its own attribute (for example emailLower), make it the key of a global secondary index and query the index. The condition in the example also makes the normalized value unique, so Alice@ and alice@ can't both register.

Tips:

• Normalize in one place and apply it to both writes and reads. toLowerCase() is locale-independent; avoid toLocaleLowerCase(), which changes letters such as the Turkish dotted i.
• Filter expressions can't lowercase a value at query time, and a filter still reads every item it evaluates. See DynamoDB filter expressions.
• For search that ignores case and accents, or matches partial words, replicate the table to Amazon OpenSearch Service with the zero-ETL integration.

Source: AWS DynamoDB documentation.