> For clean Markdown of any page, append .md to the page URL.
> For a complete documentation index, see https://docs.nvidia.com/sdgm/llms.txt.
> For AI client integration (Claude Code, Cursor, etc.), connect to the MCP server at https://docs.nvidia.com/sdgm/_mcp/server.

# kumoai

> Root module — initialization, logging, and core type enumerations

The `kumoai` root module provides SDK initialization, logging configuration, and the core type enumerations (`Dtype`, `Stype`) used throughout the SDK.

***

## `init()`

Initializes and authenticates the SDK against the Kumo service. Must be called before using any other SDK functionality.

```python
import kumoai

kumoai.init(url="<api_url>", api_key="<api_key>")
```

The Kumo API endpoint. Can also be provided via the `KUMO_API_ENDPOINT` environment variable. Inferred from the API key if not specified.

The Kumo API key. Can also be provided via the `KUMO_API_KEY` environment variable.

Snowflake credentials for SPCS deployments. Either password-based `{"user", "password", "account"}` or key-pair `{"user", "private_key", "account"}` with optional `"private_key_passphrase"` for encrypted keys. The private key must be in PEM format.

The Snowflake application name.

Logging verbosity. See [`set_log_level()`](#set-log-level) for valid values. Can also be set via the `KUMOAI_LOG` environment variable.

The group to operate in. Only applicable when RBAC is enabled. Accepts a group name or ID. Auto-selected if the user belongs to exactly one group.

The project to operate in. Only applicable when RBAC is enabled. Requires `group` to be resolvable.

***

## `set_log_level()`

Sets the logging verbosity for the Kumo SDK.

```python
kumoai.set_log_level("DEBUG")
```

The logging level. Valid values from most to least verbose: DEBUG, INFO, WARNING, ERROR, CRITICAL (`FATAL` is an alias for `CRITICAL`).

***

## `Dtype`

Enumerates the data types supported by Kumo columns.

| Value         | Description                           |
| ------------- | ------------------------------------- |
| `bool`        | Boolean column                        |
| `int`         | Integer column                        |
| `byte`        | 8-bit integer column                  |
| `int16`       | 16-bit integer column                 |
| `int32`       | 32-bit integer column                 |
| `int64`       | 64-bit integer column                 |
| `float`       | Floating-point column                 |
| `float32`     | 32-bit float column                   |
| `float64`     | 64-bit float column                   |
| `string`      | String column                         |
| `binary`      | Binary column                         |
| `date`        | Date column                           |
| `time`        | Time column                           |
| `timedelta`   | Timedelta column                      |
| `floatlist`   | List of floats (sequence / embedding) |
| `intlist`     | List of integers                      |
| `stringlist`  | List of strings                       |
| `unsupported` | Unsupported type                      |

***

## `Stype`

Enumerates the semantic types supported by Kumo columns. Semantic types describe how Kumo interprets and encodes the data in a column.

| Value              | Description                                                                                             |
| ------------------ | ------------------------------------------------------------------------------------------------------- |
| `numerical`        | A numerical column. Typically integers or floats.                                                       |
| `categorical`      | A categorical column. Typically a boolean or short string value.                                        |
| `multicategorical` | A multi-categorical column. Typically a string containing multiple concatenated category labels.        |
| `ID`               | A column holding entity identifiers.                                                                    |
| `text`             | A text column. Multi-token string values where the language content carries semantic meaning.           |
| `timestamp`        | A date/time column.                                                                                     |
| `sequence`         | A sequence or embedding column. Lists of floats of equal length, typically the output of another model. |
| `image`            | A column holding image URLs.                                                                            |
| `unsupported`      | Unsupported type.                                                                                       |