This repository contains a Python code base with best practices designed to support your MLOps initiatives.
The package leverages several tools and tips to make your MLOps experience as flexible, robust, productive as possible.
You can use this package as part of your MLOps toolkit or platform (e.g., Model Registry, Experiment Tracking, Realtime Inference, ...).
NEW: Check the MLOps Coding Course to learn how to create and use this package for your projects: https://github.com/MLOps-Courses/mlops-coding-course.
- MLOps Python Package
- Table of Contents
- Install
- Usage
- Tools
- Tips
- Resources
This section details the requirements, actions, and next steps to kickstart your MLOps project.
- Python>=3.12: to benefit from the latest features and performance improvements
- Poetry>=1.8.2: to initialize the project virtual environment and its dependencies
- Clone this GitHub repository on your computer
# with ssh (recommended)
$ git clone [email protected]:fmind/mlops-python-package
# with https
$ git clone https://github.com/fmind/mlops-python-package
$ cd mlops-python-package/
$ poetry install
- Adapt the code base to your desire
Going from there, there are dozens of ways to integrate this package to your MLOps platform.
For instance, you can use Databricks or AWS as your compute platform and model registry.
It's up to you to adapt the package code to the solution you target. Good luck champ!
This section explains how configure the project code and execute it on your system.
You can add or edit config files in the confs/
folder to change the program behavior.
# confs/training.yaml
job:
KIND: TrainingJob
inputs:
KIND: ParquetReader
path: data/inputs_train.parquet
targets:
KIND: ParquetReader
path: data/targets_train.parquet
This config file instructs the program to start a TrainingJob
with 2 parameters:
inputs
: dataset that contains the model inputstargets
: dataset that contains the model target
You can find all the parameters of your program in the src/[package]/jobs/*.py
files.
You can also print the full schema supported by this package using poetry run bikes --schema
.
The project code can be executed with poetry during your development:
$ poetry run [package] confs/tuning.yaml
$ poetry run [package] confs/training.yaml
$ poetry run [package] confs/promotion.yaml
$ poetry run [package] confs/inference.yaml
$ poetry run [package] confs/evaluations.yaml
$ poetry run [package] confs/explanations.yaml
In production, you can build, ship, and run the project as a Python package:
poetry build
poetry publish # optional
python -m pip install [package]
[package] confs/inference.yaml
You can also install and use this package as a library for another AI/ML project:
from [package] import jobs
job = jobs.TrainingJob(...)
with job as runner:
runner.run()
Additional tips:
- You can pass extra configs from the command line using the
--extras
flag- Use it to pass runtime values (e.g., a result from previous job executions)
- You can pass several config files in the command-line to merge them from left to right
- You can define common configurations shared between jobs (e.g., model params)
- The right job task will be selected automatically thanks to Pydantic Discriminated Unions
- This is a great way to run any job supported by the application (training, tuning, ....
This project includes several automation tasks to easily repeat common actions.
You can invoke the actions from the command-line or VS Code extension.
# execute the project DAG
$ inv projects
# create a code archive
$ inv packages
# list other actions
$ inv --list
Available tasks:
- checks.all (checks) - Run all check tasks.
- checks.code - Check the codes with ruff.
- checks.coverage - Check the coverage with coverage.
- checks.format - Check the formats with ruff.
- checks.poetry - Check poetry config files.
- checks.security - Check the security with bandit.
- checks.test - Check the tests with pytest.
- checks.type - Check the types with mypy.
- cleans.all (cleans) - Run all tools and folders tasks.
- cleans.cache - Clean the cache folder.
- cleans.coverage - Clean the coverage tool.
- cleans.dist - Clean the dist folder.
- cleans.docs - Clean the docs folder.
- cleans.environment - Clean the project environment file.
- cleans.folders - Run all folders tasks.
- cleans.mlruns - Clean the mlruns folder.
- cleans.mypy - Clean the mypy tool.
- cleans.outputs - Clean the outputs folder.
- cleans.poetry - Clean poetry lock file.
- cleans.pytest - Clean the pytest tool.
- cleans.projects - Run all projects tasks.
- cleans.python - Clean python caches and bytecodes.
- cleans.requirements - Clean the project requirements file.
- cleans.reset - Run all tools, folders, and sources tasks.
- cleans.ruff - Clean the ruff tool.
- cleans.sources - Run all sources tasks.
- cleans.tools - Run all tools tasks.
- cleans.venv - Clean the venv folder.
- commits.all (commits) - Run all commit tasks.
- commits.bump - Bump the version of the package.
- commits.commit - Commit all changes with a message.
- commits.info - Print a guide for messages.
- containers.all (containers) - Run all container tasks.
- containers.build - Build the container image with the given tag.
- containers.compose - Start up docker compose.
- containers.run - Run the container image with the given tag.
- docs.all (docs) - Run all docs tasks.
- docs.api - Document the API with pdoc using the given format and output directory.
- docs.serve - Serve the API docs with pdoc using the given format and computer port.
- formats.all - (formats) Run all format tasks.
- formats.imports - Format python imports with ruff.
- formats.sources - Format python sources with ruff.
- installs.all (installs) - Run all install tasks.
- installs.poetry - Install poetry packages.
- installs.pre-commit - Install pre-commit hooks on git.
- mlflow.all (mlflow) - Run all mlflow tasks.
- mlflow.doctor - Run mlflow doctor to diagnose issues.
- mlflow.serve - Start mlflow server with the given host, port, and backend uri.
- packages.all (packages) - Run all package tasks.
- packages.build - Build a python package with the given format.
- projects.all (projects) - Run all project tasks.
- projects.environment - Export the project environment file.
- projects.requirements - Export the project requirements file.
- projects.run - Run an mlflow project from MLproject file.
This package supports two GitHub Workflows in .github/workflows
:
check.yml
: validate the quality of the package on each Pull Requestpublish.yml
: build and publish the docs and packages on code release.
You can use and extend these workflows to automate repetitive package management tasks.
This sections motivates the use of developer tools to improve your coding experience.
Pre-defined actions to automate your project development.
Commits: Commitizen
- Motivations:
- Limitations:
- Learning curve for new users
- Alternatives:
- Do It Yourself (DIY)
Git Hooks: Pre-Commit
- Motivations:
- Check your code locally before a commit
- Avoid wasting resources on your CI/CD
- Can perform extra (e.g., file cleanup)
- Limitations:
- Add overhead before your commit
- Alternatives:
- Git Hooks: less convenient to use
Tasks: PyInvoke
- Motivations:
- Automate project workflows
- Sane syntax compared to alternatives
- Good trade-off between power/simplicity
- Limitations:
- Not familiar to most developers
- Alternatives:
- Make: most popular, but awful syntax
Execution of automated workflows on code push and releases.
Runner: GitHub Actions
- Motivations:
- Native on GitHub
- Simple workflow syntax
- Lots of configs if needed
- Limitations:
- SaaS Service
- Alternatives:
- GitLab: can be installed on-premise
Integrations with the Command-Line Interface (CLI) of your system.
Parser: Argparse
- Motivations:
- Provide CLI arguments
- Included in Python runtime
- Sufficient for providing configs
- Limitations:
- More verbose for advanced parsing
- Alternatives:
Logging: Loguru
- Motivations:
- Show progress to the user
- Work fine out of the box
- Saner logging syntax
- Limitations:
- Doesn't let you deviate from the base usage
- Alternatives:
- Logging: available by default, but feel dated
Edition, validation, and versioning of your project source code.
Coverage: Coverage
- Motivations:
- Report code covered by tests
- Identify code path to test
- Show maturity to users
- Limitations:
- None
- Alternatives:
- None?
Editor: VS Code
- Motivations:
- Open source
- Free, simple, open source
- Great plugins for Python development
- Limitations:
- Require some configuration for Python
- Alternatives:
Formatting: Ruff
- Motivations:
- Super fast compared to others
- Don't waste time arranging your code
- Make your code more readable/maintainable
- Limitations:
- Still in version 0.x, but more and more adopted
- Alternatives:
Quality: Ruff
- Motivations:
- Improve your code quality
- Super fast compared to others
- Great integration with VS Code
- Limitations:
- None
- Alternatives:
Security: Bandit
- Motivations:
- Detect security issues
- Complement linting solutions
- Not to heavy to use and enable
- Limitations:
- None
- Alternatives:
- None
Testing: Pytest
- Motivations:
- Write tests or pay the price
- Super easy to write new test cases
- Tons of good plugins (xdist, sugar, cov, ...)
- Limitations:
- Doesn't support parallel execution out of the box
- Alternatives:
- Unittest: more verbose, less fun
Typing: Mypy
- Motivations:
- Static typing is cool!
- Communicate types to use
- Official type checker for Python
- Limitations:
- Can have overhead for complex typing
- Alternatives:
Versioning: Git
- Motivations:
- If you don't version your code, you are a fool
- Most popular source code manager (what else?)
- Provide hooks to perform automation on some events
- Limitations:
- Git can be hard: https://xkcd.com/1597/
- Alternatives:
- Mercurial: loved it back then, but git is the only real option
Manage the configs files of your project to change executions.
Format: YAML
- Motivations:
- Change execution without changing code
- Readable syntax, support comments
- Allow to use OmegaConf <3
- Limitations:
- Not supported out of the box by Python
- Alternatives:
Parser: OmegaConf
- Motivations:
- Parse and merge YAML files
- Powerful, doesn't get in your way
- Achieve a lot with few lines of code
- Limitations:
- Do not support remote files (e.g., s3, gcs, ...)
- You can combine it with cloudpathlib
- Do not support remote files (e.g., s3, gcs, ...)
- Alternatives:
Reader: Cloudpathlib
- Motivations:
- Read files from cloud storage
- Better integration with cloud platforms
- Support several platforms: AWS, GCP, and Azure
- Limitations:
- Support of Python typing is not great at the moment
- Alternatives:
- Cloud SDK (GCP, AWS, Azure, ...): vendor specific, overkill for this task
Validator: Pydantic
- Motivations:
- Validate your config before execution
- Pydantic should be builtin (period)
- Super charge your Python class
- Limitations:
- None
- Alternatives:
Define the datasets to provide data inputs and outputs.
Container: Pandas
- Motivations:
- Load data files in memory
- Lingua franca for Python
- Most popular options
- Limitations:
- Lot of gotchas
- Alternatives:
Format: Parquet
- Motivations:
- Store your data on disk
- Column-oriented (good for analysis)
- Much more efficient and saner than text based
- Limitations:
- None
- Alternatives:
Schema: Pandera
- Motivations:
- Typing for dataframe
- Communicate data fields
- Support pandas and others
- Limitations:
- None
- Alternatives:
- Great Expectations: powerful, but much more difficult to integrate
Generate and share the project documentations.
API: pdoc
- Motivations:
- Share docs with others
- Simple tool, only does API docs
- Get the job done, get out of your way
- Limitations:
- Only support API docs (i.e., no custom docs)
- Alternatives:
Format: Google
- Motivations:
- Common style for docstrings
- Most writeable out of alternatives
- I often write a single line for simplicity
- Limitations:
- None
- Alternatives:
Hosting: GitHub Pages
- Motivations:
- Easy to setup
- Free and simple
- Integrated with GitHub
- Limitations:
- Only support static content
- Alternatives:
- ReadTheDocs: provide more features
Toolkit to handle machine learning models.
Evaluation: Scikit-Learn Metrics
- Motivations:
- Bring common metrics
- Avoid reinventing the wheel
- Avoid implementation mistakes
- Limitations:
- Limited set of metric to be chosen
- Alternatives:
- Implement your own: for custom metrics
Format: Mlflow Model
- Motivations:
- Standard ML format
- Store model dependencies
- Strong community ecosystem
- Limitations:
- None
- Alternatives:
- Pickle: work out of the box, but less suited for big array
- ONNX: great for deep learning, no guaranteed compatibility for the rest
Registry: Mlflow Registry
- Motivations:
- Save and load models
- Separate production from consumption
- Popular, open source, work on local system
- Limitations:
- None
- Alternatives:
- Neptune.ai: SaaS solution
- Weights and Biases: SaaS solution
Tracking: Mlflow Tracking
- Motivations:
- Keep track of metrics and params
- Allow to compare model performances
- Popular, open source, work on local system
- Limitations:
- None
- Alternatives:
- Neptune.ai: SaaS solution
- Weights and Biases: SaaS solution
Define and build modern Python package.
Evolution: Changelog
- Motivation:
- Communicate changes to user
- Can be updated with Commitizen
- Standardized with Keep a Changelog
- Limitations:
- None
- Alternatives:
- None
Format: Wheel
- Motivations:
- Has several advantages
- Create source code archive
- Most modern Python format
- Limitations:
- Doesn't ship with C/C dependencies (e.g., CUDA)
- i.e., use Docker containers for this case
- Doesn't ship with C/C dependencies (e.g., CUDA)
- Alternatives:
Manager: Poetry
- Motivations:
- Define and build Python package
- Most popular solution by GitHub stars
- Pack every metadata in a single static file
- Limitations:
- Cannot add dependencies beyond Python (e.g., CUDA)
- i.e., use Docker container for this use case
- Cannot add dependencies beyond Python (e.g., CUDA)
- Alternatives:
- Setuptools: dynamic file is slower and more risky
- Pdm, Hatch, PipEnv: https://xkcd.com/1987/
Runtime: Docker
- Motivations:
- Create isolated runtime
- Container is the de facto standard
- Package C/C dependencies with your project
- Limitations:
- Some company might block Docker Desktop, you should use alternatives
- Alternatives:
- Conda: slow and heavy resolver
Select your programming environment.
Language: Python
- Motivations:
- Great language for AI/ML projects
- Robust with additional tools
- Hundreds of great libs
- Limitations:
- Slow without C bindings
- Alternatives:
Version: Pyenv
- Motivations:
- Switch between Python version
- Allow to select the best version
- Support global and local dispatch
- Limitations:
- Require some shell configurations
- Alternatives:
- Manual installation: time consuming
Reproducibility: Mlflow Project
- Motivations:
- Share common project formats.
- Ensure the project can be reused.
- Avoid randomness in project execution.
- Limitations:
- Mlflow Project is best suited for small projects.
- Alternatives:
- DVC: both data and models.
- Metaflow: focus on machine learning.
- Apache Airflow: for large scale projects.
Monitoring : Mlflow Evaluate
- Motivations:
- Compute the model metrics.
- Validate model with thresholds.
- Perform post-training evaluations.
- Limitations:
- Mlflow Evaluate is less feature-rich as alternatives.
- Alternatives:
Alerting: Plyer
- Motivations:
- Simple solution.
- Send notifications on system.
- Cross-system: Mac, Linux, Windows.
- Limitations:
- Should not be used for large scale projects.
- Alternatives:
Lineage: Mlflow Dataset
- Motivations:
- Store information in Mlflow.
- Track metadata about run datasets.
- Keep URI of the dataset source (e.g., website).
- Limitations:
- Not as feature-rich as alternative solutions.
- Alternatives:
- Databricks Lineage: limited to Databricks.
- OpenLineage and Marquez: open-source and flexible.
Explainability: SHAP
- Motivations:
- Most popular toolkit.
- Support various models (linear, model, ...).
- Integration with Mlflow through the SHAP module.
- Limitations:
- Super slow on large dataset.
- Mlflow SHAP module is not mature enough.
- Alternatives:
- LIME: not maintained anymore.
Infrastructure: Mlflow System Metrics
- Motivations:
- Track infrastructure information (RAM, CPU, ...).
- Integrated with Mlflow tracking.
- Provide hardware insights.
- Limitations:
- Not as mature as alternative solutions.
- Alternatives:
- Datadog: popular and mature solution.
This sections gives some tips and tricks to enrich the develop experience.
You should decouple the pointer to your data from how to access it.
In your code, you can refer to your dataset with a tag (e.g., inputs
, targets
).
This tag can then be associated to a reader/writer implementation in a configuration file:
inputs:
KIND: ParquetReader
path: data/inputs_train.parquet
targets:
KIND: ParquetReader
path: data/targets_train.parquet
In this package, the implementation are described in src/[package]/io/datasets.py
and selected by KIND
.
You should select the best hyperparameters for your model using optimization search.
The simplest projects can use a sklearn.model_selection.GridSearchCV
to scan the whole search space.
This package provides a simple interface to this hyperparameter search facility in src/[package]/utils/searchers.py
.
For more complex project, we recommend to use more complex strategy (e.g., Bayesian) and software package (e.g., Optuna).
You should properly split your dataset into a training, validation, and testing sets.
- Training: used for fitting the model parameters
- Validation: used to find the best hyperparameters
- Testing: used to evaluate the final model performance
The sets should be exclusive, and the testing set should never be used as training inputs!
This package provides a simple deterministic strategy implemented in src/[package]/utils/splitters.py
.
You should use Directed-Acyclic Graph (DAG) to connect the steps of your ML pipeline.
A DAG can express the dependencies between steps while keeping the individual step independent.
This package provides a simple DAG example in tasks/dags.py
. This approach is based on PyInvoke.
In production, we recommend to use a scalable system such as Airflow, Dagster, Prefect, Metaflow, or ZenML.
You should provide a global context for the execution of your program.
There are several approaches such as Singleton, Global Variable, or Component.
This package takes inspiration from Clojure mount. It provides an implementation in src/[package]/io/services.py
.
You should separate the program implementation from the program configuration.
Exposing configurations to users allow them to influence the execution behavior without code changes.
This package seeks to expose as much parameter as possible to the users in configurations stored in the confs/
folder.
You should implement the SOLID principles to make your code as flexible as possible.
- Single responsibility principle: Class has one job to do. Each change in requirements can be done by changing just one class.
- Open/closed principle: Class is happy (open) to be used by others. Class is not happy (closed) to be changed by others.
- Liskov substitution principle: Class can be replaced by any of its children. Children classes inherit parent's behaviours.
- Interface segregation principle: When classes promise each other something, they should separate these promises (interfaces) into many small promises, so it's easier to understand.
- Dependency inversion principle: When classes talk to each other in a very specific way, they both depend on each other to never change. Instead classes should use promises (interfaces, parents), so classes can change as long as they keep the promise.
In practice, this mean you can implement software contracts with interface and swap the implementation.
For instance, you can implement several jobs in src/[package]/jobs/*.py
and swap them in your configuration.
To learn more about the mechanism select for this package, you can check the documentation for Pydantic Tagged Unions.
You should separate the code interacting with the external world from the rest.
The external is messy and full of risks: missing files, permission issue, out of disk ...
To isolate these risks, you can put all the related code in an io
package and use interfaces
You should use Python context manager to control and enhance an execution.
Python provides contexts that can be used to extend a code block. For instance:
# in src/[package]/scripts.py
with job as runner: # context
runner.run() # run in context
This pattern has the same benefit as Monad, a powerful programming pattern.
The package uses src/[package]/jobs/*.py
to handle exception and services.
You should create Python package to create both library and application for others.
Using Python package for your AI/ML project has the following benefits:
- Build code archive (i.e., wheel) that be uploaded to Pypi.org
- Install Python package as a library (e.g., like pandas)
- Expose script entry points to run a CLI or a GUI
To build a Python package with Poetry, you simply have to type in a terminal:
# for all poetry project
poetry build
# for this project only
inv packages
You should type your Python code to make it more robust and explicit for your user.
Python provides the typing module for adding type hints and mypy to checking them.
# in src/[package]/core/models.py
@abc.abstractmethod
def fit(self, inputs: schemas.Inputs, targets: schemas.Targets) -> "Model":
"""Fit the model on the given inputs and target."""
@abc.abstractmethod
def predict(self, inputs: schemas.Inputs) -> schemas.Outputs:
"""Generate an output with the model for the given inputs."""
This code snippet clearly state the inputs and outputs of the method, both for the developer and the type checker.
The package aims to type every functions and classes to facilitate the developer experience and fix mistakes before execution.
You should type your configuration to avoid exceptions during the program execution.
Pydantic allows to define classes that can validate your configs during the program startup.
# in src/[package]/utils/splitters.py
class TrainTestSplitter(Splitter):
shuffle: bool = False # required (time sensitive)
test_size: int | float = 24 * 30 * 2 # 2 months
random_state: int = 42
This code snippet allows to communicate the values expected and avoid error that could be avoided.
The package combines both OmegaConf and Pydantic to parse YAML files and validate them as soon as possible.
You should type your dataframe to communicate and validate their fields.
Pandera supports dataframe typing for Pandas and other library like PySpark:
# in src/package/schemas.py
class InputsSchema(Schema):
instant: papd.Index[papd.UInt32] = pa.Field(ge=0, check_name=True)
dteday: papd.Series[papd.DateTime] = pa.Field()
season: papd.Series[papd.UInt8] = pa.Field(isin=[1, 2, 3, 4])
yr: papd.Series[papd.UInt8] = pa.Field(ge=0, le=1)
mnth: papd.Series[papd.UInt8] = pa.Field(ge=1, le=12)
hr: papd.Series[papd.UInt8] = pa.Field(ge=0, le=23)
holiday: papd.Series[papd.Bool] = pa.Field()
weekday: papd.Series[papd.UInt8] = pa.Field(ge=0, le=6)
workingday: papd.Series[papd.Bool] = pa.Field()
weathersit: papd.Series[papd.UInt8] = pa.Field(ge=1, le=4)
temp: papd.Series[papd.Float16] = pa.Field(ge=0, le=1)
atemp: papd.Series[papd.Float16] = pa.Field(ge=0, le=1)
hum: papd.Series[papd.Float16] = pa.Field(ge=0, le=1)
windspeed: papd.Series[papd.Float16] = pa.Field(ge=0, le=1)
casual: papd.Series[papd.UInt32] = pa.Field(ge=0)
registered: papd.Series[papd.UInt32] = pa.Field(ge=0)
This code snippet defines the fields of the dataframe and some of its constraint.
The package encourages to type every dataframe used in src/[package]/core/schemas.py
.
You should use the Objected Oriented programming to benefit from polymorphism.
Polymorphism combined with SOLID Principles allows to easily swap your code components.
class Reader(abc.ABC, pdt.BaseModel):
@abc.abstractmethod
def read(self) -> pd.DataFrame:
"""Read a dataframe from a dataset."""
This code snippet uses the abc module to define code interfaces for a dataset with a read/write method.
The package defines class interface whenever possible to provide intuitive and replaceable parts for your AI/ML project.
You should use semantic versioning to communicate the level of compatibility of your releases.
Semantic Versioning (SemVer) provides a simple schema to communicate code changes. For package X.Y.Z:
- Major (X): major release with breaking changed (i.e., imply actions from the benefit)
- Minor (Y): minor release with new features (i.e., provide new capabilities)
- Patch (Z): patch release to fix bugs (i.e., correct wrong behavior)
Poetry and this package leverage Semantic Versioning to let developers control the speed of adoption for new releases.
You can run your tests in parallel to speed up the validation of your code base.
Pytest can be extended with the pytest-xdist plugin for this purpose.
This package enables Pytest in its automation tasks by default.
You should define reusable objects and actions for your tests with fixtures.
Fixture can prepare objects for your test cases, such as dataframes, models, files.
This package defines fixtures in tests/conftest.py
to improve your testing experience.
You can use VS Code workspace to define configurations for your project.
Code Workspace can enable features (e.g. formatting) and set the default interpreter.
{
"settings": {
"editor.formatOnSave": true,
"python.defaultInterpreterPath": ".venv/bin/python",
...
},
}
This package defines a workspace file that you can load from [package].code-workspace
.
You can use GitHub Copilot to increase your coding productivity by 30%.
GitHub Copilot has been a huge productivity thanks to its smart completion.
You should become familiar with the solution in less than a single coding session.
You can use VIM keybindings to more efficiently navigate and modify your code.
Learning VIM is one of the best investment for a career in IT. It can make you 30% more productive.
Compared to GitHub Copilot, VIM can take much more time to master. You can expect a ROI in less than a month.
This section provides resources for building packages for Python and AI/ML/MLOps.