Clarity before complexity
The workflow is for people who want to understand what happened, not just receive a score. Every step of BreezeML can be read, printed, and questioned.
The maker
Akash Anipakalu Giridhar
Solutions Engineer. Creator of BreezeML.
BreezeML is a free, open-source Python library I built on top of scikit-learn, released under the MIT licence. You give it a dataframe and the name of the target column; it detects the task, preprocesses the data, cross-validates, and returns a model.
The rest of the library exists to check that model before you trust it: leakage detection, fairness gaps measured across groups, significance tests between two models, conformal prediction intervals, drift monitoring, and a generated model card. export() writes standalone scikit-learn code, so you can stop using BreezeML and keep the model.
The core installs with four dependencies, scikit-learn, pandas, numpy and joblib, and a CI test fails the build if a fifth is added.
I designed the library, wrote the code, and maintain it along with this website and its documentation.
The principles the library is built on
Honesty before optimization
Leakage, fairness gaps, drift and uncertainty should surface on your laptop, not in production. The library tells you the uncomfortable thing first.
Leave freely
export() writes standalone scikit-learn code. A tool you cannot leave is a tool you cannot trust, so BreezeML is built to be walked away from.
BreezeML at a glance
v2.0.0
latest PyPI release
55
commits in the main repo
2
GitHub contributors
3
forks
Find the work
The release series
All releases →Honest by Default, Agent-native by Design
report() one-verdict scorecard (SHIP/WARN/STOP), unified Model object, conformal predict_interval/predict_set, py.typed, MCP report tool
Honest Uncertainty and Cause
conformal prediction, active learning, auto feature engineering, causal inference
Four Questions
significance testing, multi-output, recommenders, survival analysis
The Honest Machine
audit, fairness, imbalance, blend, track, anomaly, semi-supervised, CLI
The Four Breaths
layered architecture and guide()
The Zen Garden
zen(), haiku(), fortune(), sensei()
Questions
- Who created BreezeML?
- BreezeML was created by Akash Anipakalu Giridhar, a Solutions Engineer. He designed the library, wrote the code, and maintains the project along with its documentation and this website.
- What does Akash Anipakalu Giridhar work on?
- He works as a Solutions Engineer and builds open-source machine-learning tooling. BreezeML, his main open-source project, wraps scikit-learn in a short workflow and adds the checks around it: data leakage detection, fairness gaps, drift, conformal prediction intervals, and generated model cards.
- Where can I find his code and profiles?
- His code is on GitHub at github.com/venomez-viper, the library ships on PyPI as breezeml, his personal portfolio is at akashanipakalugiridhar.com, and he is on LinkedIn as Akash Anipakalu Giridhar.