Skip to main content

Uncertainty Engine banner

Python SDK for the Uncertainty Engine

PyPI Python Versions

⚠️ Pre-Release Notice: This SDK is currently in pre-release development. Please ensure you are reading documentation that corresponds to the specific version of the SDK you have installed, as features and APIs may change between versions.

Requirements

  • Python >=3.10, <3.13
  • Valid Uncertainty Engine account

Installation

pip install uncertainty-engine

The SDK has optional extras that provide additional functionality:

Name Description
data Installs data processing packages
notebook Installs support for running the SDK in Jupyter notebooks
vis Installs support for visualising graphs

Install one or more of these via pip by providing them as a comma-separated list:

pip install "uncertainty-engine[data,notebook,vis]"

Usage

Setting your username and password

To run and queue workflows you must have your Uncertainty Engine username and password set up. To do this you can run the following in your terminal:

export UE_USERNAME="your_username"
export UE_PASSWORD="your_password"

Creating a client

All interactions with the Uncertainty Engine API are performed via a Client. The client can be defined as follows:

from uncertainty_engine import Client

client = Client()

With an instantiated Client object, and username and password set as environmental variables, authentication can be carried via the following:

client.authenticate()

Using different environments

The Client defaults to using the Uncertainty Engine production environment. To use a different named environment, pass it as the env argument:

client = Client(env="dev")

If a custom environment has been provided for you, pass an Environment that describes the details:

from uncertainty_engine import Client, Environment

client = Client(
    env=Environment(
        cognito_user_pool_client_id="…",
        core_api="…",
        region="…",
        resource_api="…",
    ),
)
Argument Format Example
cognito_user_pool_client_id Alphanumeric string 3vj5pe253j4v070euqjdk38a24
core_api Starts with https://, does not end with / https://de1v75vvk6.execute-api.eu-west-2.amazonaws.com
region Geographic region code eu-west-2
resource_api Starts with https://, does not end with / https://m90q55iux6.execute-api.eu-west-2.amazonaws.com

Note: Every password is tied to a specific environment. A password for the production environment, for example, won't grant access to the development environment. Ensure you set the correct UE_PASSWORD value for the environment you configure.

Troubleshooting

Authorisation tokens are cached in .ue_auth in your home directory. If the SDK fails to authenticate you (for example, after switching from one environment to another), delete .ue_auth to generate and cache new tokens.

To delete the cache in macOS or Linux:

rm ~/.ue_auth

To delete the cache in Windows:

del "%USERPROFILE%\.ue_auth"

Building nodes

Every node registered with the Uncertainty Engine can be built through client.nodes. There are no per-node classes to import: the SDK looks each node up when you use it, so new nodes, and nodes you register yourself, work straight away.

Finding nodes

client.nodes.available()              # the names of every node you can use
client.nodes.describe("Add")          # a node's description, inputs and outputs
client.nodes.describe("Add").inputs   # just the inputs it takes
client.nodes.describe("Add").outputs  # and what it produces

In Jupyter or IPython you can also type client.nodes. and press Tab to autocomplete node names.

Building and running a node

from pprint import pprint

add = client.nodes.Add(lhs=1, rhs=2, label="add")

response = client.run_node(add)
pprint(response.outputs)

If the node's name is in a variable, call client.nodes with it instead: client.nodes("Add", lhs=1, rhs=2, label="add").

Inputs are checked against the node's schema as the node is built. A missing or unknown input raises a NodeValidationError, and a node that doesn't exist raises a NodeNotFoundError; both are in uncertainty_engine.exceptions. Inputs are passed exactly as the node declares them, so a node that takes a resource ID, for example, expects {"id": ...}.

Versions

By default a node is built at the version the Uncertainty Engine considers its default: "latest" if there is one, otherwise the highest version. To build a specific version, pass version, or pin several nodes at once with with_versions():

add = client.nodes.Add(lhs=1, rhs=2, label="add", version="0.2.0")

pinned = client.nodes.with_versions({"TrainModel": "0.2.0", "PredictModel": "0.2.0"})
train = pinned.TrainModel(...)  # built at 0.2.0; nodes not in the mapping use their default

client.get_node_versions("Add") lists the versions available.

Caching

A node's schema is fetched the first time you build it, then kept for the life of the client, as is the list from available(). To pick up a node or version deployed since, clear the cache:

client.clear_node_cache()

See examples/dynamic_nodes.ipynb for a full walkthrough, and our example notebooks for more in-depth examples.

Upgrading from node classes

Earlier versions of the SDK had a Python class for each node, such as from uncertainty_engine.nodes.basic import Add. Those classes have been removed in favour of client.nodes; see MIGRATION.md for how to update your code.

Metadata

Release files for uncertainty-engine 0.15.0

For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.

Source distribution (sdist)

Source distribution for uncertainty-engine 0.15.0
File Size Uploaded Provenance Actions
uncertainty_engine-0.15.0.tar.gz 41.4 kB Without provenance Details for uncertainty_engine-0.15.0.tar.gz

Built distribution (wheel)

Table of built distributions (wheels) for uncertainty-engine 0.15.0
File Interpreter ABI Platform Actions
uncertainty_engine-0.15.0-py3-none-any.whl Python 3 none any Details for uncertainty_engine-0.15.0-py3-none-any.whl

Total release size: 97.8 kB

Release files / uncertainty_engine-0.15.0.tar.gz

Download URL uncertainty_engine-0.15.0.tar.gz
Size 41.4 kB
Tags Source
SHA-256 checksum
How to use checksums
fef8adea9f393ec1125d0808ce3c883f875e25c0bc2bb62883659c9a50982a2b
BLAKE2b-256 checksum
How to use checksums
0d02f70ff6d7d5cc8a87d2f9658b1ce32ea6765a244628ff3bddbe0b672a527d
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via poetry/2.1.4 CPython/3.13.5 Darwin/25.3.0

Release files / uncertainty_engine-0.15.0-py3-none-any.whl

Download URL uncertainty_engine-0.15.0-py3-none-any.whl
Size 56.4 kB
Tags Python 3
SHA-256 checksum
How to use checksums
2e4457bba5ccb3804fc2d3c92465e85d17a0d188daa74ab4ef561bcd004d4790
BLAKE2b-256 checksum
How to use checksums
3dcea9a5b047cac5ec92f4a7239dc9543a9189cfd151a4d14a8a599cb0b78b6f
Upload date
Uploaded using Trusted Publishing?
What is trusted publishing?
No
Uploaded via poetry/2.1.4 CPython/3.13.5 Darwin/25.3.0

Security for uncertainty-engine 0.15.0

Provenance

Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.

No provenance found

None of the files in this release were uploaded with provenance metadata.

A note on package security

PyPI is a repository for distributing packages, not a service for auditing them. While third-party security researchers and our staff do their best to detect and remove malicious packages, PyPI cannot guarantee the quality, security or suitability of any given package.

To learn more about protecting yourself from malicious actors, visit our package security documentation.

Anthropic, PBC Visionary sponsor Bloomberg Visionary sponsor Hudson River Trading Visionary sponsor Meta Visionary sponsor NVIDIA Visionary sponsor Microsoft Sustainability sponsor Depot Continuous Integration AWS Cloud computing and Security Sponsor Datadog Monitoring Fastly CDN Google Download Analytics Sentry Error logging StatusPage Status page