init-databricks-ml

Initialize / Update Databricks ML Project

Set up or update the current project with Databricks + MLflow + Claude Code integration. For each item below, CHECK if it already exists before creating. If it exists, skip it or merge missing parts. Only create/update what is missing.

Checklist

1. .gitignore

Check if .gitignore exists. If missing, create it with:

# Python-generated files
__pycache__/
*.py[oc]
*.py[cod]
build/
dist/
wheels/
*.egg-info
*.whl

# Virtual environments
.venv/
venv/

# Environment
.env*
!.env.example

# MLflow
mlflow_results/

# EDA / Research
eda_results/
research/

# OS
.DS_Store

# Claude
.claude/

# uv
uv.lock

# misc
std.out

If it exists, check that .env*, !.env.example, mlflow_results/, .venv/, eda_results/, research/, .claude/, uv.lock, and std.out are listed. Append missing entries.

1b. .claudeignore

Check if .claudeignore exists. If missing, create it. This file prevents Claude Code from indexing large data files, model artifacts, and binary files that are irrelevant to code understanding.

# Data files — large binary/tabular data that Claude doesn't need to read
*.csv
*.tsv
*.parquet
*.feather
*.arrow
*.h5
*.hdf5
*.pkl
*.pickle
*.npy
*.npz
*.zarr
*.tfrecord
*.avro
*.orc

# Data directories
data/
datasets/

# Model artifacts
*.pt
*.pth
*.onnx
*.safetensors
*.bin
*.ckpt
models/

# Images / media
*.png
*.jpg
*.jpeg
*.gif
*.bmp
*.svg
*.mp4
*.mp3
*.wav

# Archives
*.zip
*.tar
*.tar.gz
*.tgz
*.gz
*.bz2
*.7z
*.rar

# Virtual environments
.venv/
venv/

# Build artifacts
build/
dist/
wheels/
*.egg-info/
*.whl

# MLflow local results
mlflow_results/

# EDA / Research outputs
eda_results/
research/

# Misc
__pycache__/
*.pyc
.DS_Store
std.out
uv.lock

If it exists, check that the data file extensions (*.csv, *.parquet, *.h5, *.pkl, *.npy), data directories (data/, datasets/), model artifact extensions (*.pt, *.pth, *.onnx, *.safetensors), and build/output directories (mlflow_results/, eda_results/, research/) are listed. Append missing entries.