📘 README.md — Pyton Package Using Conda

A fully self‑healing, auto-fixing, auto‑versioned, auto‑releasing MLOps system.

 

🚀 Overview

OASIS is a fully autonomous Machine Learning + DevOps hybrid pipeline featuring:

Real‑dataset LightGBM training

Versioned model saving

Semantic versioning

Full CLI toolkit ( oasis train ,  oasis version ,  oasis auto‑fix , etc.)

Automatic changelog generation

Automatic GitHub Releases

CI Retry + Auto‑Merge system

PR‑based self‑healing

Auto‑close failing PRs

Nightly auto‑fix pipelines

Auto‑formatting, linting, diagnostics, and repository cleanup

OASIS maintains itself — heals its own repo, fixes CI failures, formats code, retries CI, publishes releases, updates changelogs, and more.

 

📁 Project Structure

  OASIS/ │ ├── data/ │ └── dataset.csv │ ├── models/ │ ├── trained_model.pkl │ ├── version.txt │ └── history.log │ ├── src/ │ ├── train_pipeline.py │ ├── model_loader.py │ └── oasis/ │ └── cli.py │ ├── tests/ │ └── test_lgb_model.py │ └── .github/workflows/ ├── ci.yml ├── oasis-auto-fix.yml ├── oasis-auto-fix-pr.yml ├── oasis-auto-fix-nightly.yml ├── oasis-auto-merge.yml ├── oasis-auto-close.yml └── oasis-ci-retry.yml  

 

🧠 Training Pipeline

Training uses:

  src/train_pipeline.py  

Pipeline includes:

Loading real dataset

Splitting training/test

Training LightGBM

Saving model + metadata

Recording semantic version

Appending version history

Train manually:

  oasis train  

 

🧪 Testing

Tests validate:

Model load

Feature alignment

Prediction behavior

Deterministic output

Run manually:

  pytest -v  

 

⚙️ GitHub Actions Overview

OASIS includes 7 fully autonomous workflows:

✔  ci.yml 

Standard train + test workflow.

✔  oasis-auto-fix.yml 

Self-heals repository on command.

✔  oasis-auto-fix-pr.yml 

Creates auto-fix PRs instead of pushing changes.

✔  oasis-auto-fix-nightly.yml 

Runs nightly repository healing at 2AM UTC.

✔  oasis-auto-merge.yml 

Auto-merges approved auto-fix PRs only when CI is green.

✔  oasis-auto-close.yml 

Auto-closes persistent failing PRs after 3 CI failures.

✔  oasis-ci-retry.yml 

Retries CI up to 3 times before merging or closing.

Combined, these workflows create a self-maintaining MLOps ecosystem.

 

🧵 OASIS CLI Commands

Your CLI includes:

🔧 Training & Model Management

  oasis train oasis evaluate <dataset.csv> oasis predict <input.csv>  

🔍 Model Metadata

  oasis version oasis version --json  

Metadata includes:

Semantic version

Timestamp

Feature list

Model size

File path

🧾 Version History & Releases

  oasis bump-version --level patch|minor|major oasis history oasis changelog oasis release  

Release automatically:

Tags Git

Generates changelog

Uploads model to GitHub Releases

🛠 Auto‑Fix & Formatting

  oasis auto-fix oasis auto-fix-strict oasis format oasis clean  

🩺 Diagnostics

  oasis doctor oasis doctor --json oasis doctor --fix oasis doctor --fix --commit --push  

Doctor checks:

Python syntax

YAML health

GPU availability

Missing dependencies

Model file integrity

Git status

Auto-healing

 

🤖 Self‑Healing DevOps Explained

OASIS includes autonomous maintenance loops:

1️⃣ Failure → Auto-Fix PR

A CI failure triggers a repair branch & PR.

2️⃣ Auto‑Retry CI

OASIS retries CI up to 3 times.

3️⃣ Auto‑Comment Failure Reasons

Explains why CI failed directly on PR.

4️⃣ Auto‑Merge

If CI passes + PR is approved → merge.

5️⃣ Auto‑Close

If CI fails 3 times → PR closed with explanation.

6️⃣ Nightly Repair

Nightly self-healing runs regardless of CI.

 

🚀 Release Automation

Release with:

  oasis release  

This:

Reads semantic version

Creates Git tag

Generates changelog

Uploads model

Publishes GitHub Release

Optional:

  oasis release --no-confirm oasis release --notes "Custom message"  

 

🧹 Cleanup & Formatting

Run:

  oasis clean oasis format  

Removes:

Caches

Build files

Logs

Model artifacts (optional)

And formats code using:

Black

isort

docformatter

 

📦 Installation

Editable mode installation:

  pip install -e .  

 

🛟 Support

If you need enhancements, improvements, or more automation, extend the CLI or GitHub workflows.

 

🎉 Final Note

This README documents your complete autonomous ML + DevOps pipeline. Your OASIS system is now capable of:

Training

Testing

Healing

Formatting

Releasing

Versioning

Closing

Commenting

Auto-merging

Nightly cleaning

all without human intervention.