Initial commit: AutoML MLOps Pipeline project structure

.gitignore

@@ -0,0 +1,83 @@
+__pycache__/
+*.py[cod]
+*$py.class
+*.so
+.Python
+build/
+develop-eggs/
+dist/
+downloads/
+eggs/
+.eggs/
+lib/
+lib64/
+parts/
+sdist/
+var/
+wheels/
+*.egg-info/
+.installed.cfg
+*.egg
+
+.env
+.venv
+env/
+venv/
+ENV/
+env.bak/
+venv.bak/
+automlenv/
+
+.pytest_cache/
+.coverage
+htmlcov/
+*.cover
+.hypothesis/
+
+*.log
+
+.vscode/
+.idea/
+*.swp
+*.swo
+*~
+
+artifacts/
+!artifacts/.gitkeep
+mlflow/mlruns/
+mlruns/
+mlartifacts/
+
+.dvc/cache/
+.dvc/tmp/
+
+data/
+!data/.gitkeep
+!data/sample_data.csv
+
+*.h5
+*.pkl
+*.joblib
+*.model
+*.onnx
+*.pb
+
+.ipynb_checkpoints/
+*.ipynb_checkpoints
+
+.DS_Store
+Thumbs.db
+*.bak
+*.tmp
+
+node_modules/
+package-lock.json
+
+.mypy_cache/
+.dmypy.json
+dmypy.json
+.pyre/
+.pytype/
+
+models/production/
+!models/.gitkeep

config/automl_config.yaml

@@ -0,0 +1,44 @@
+automl_library: autogluon
+
+autogluon:
+  time_limit: 600
+  presets: medium_quality
+  eval_metric: null
+  verbosity: 2
+  num_bag_folds: 5
+  num_stack_levels: 1
+
+flaml:
+  time_budget: 600
+  metric: auto
+  task: classification
+  estimator_list:
+    - lgbm
+    - xgboost
+    - rf
+    - extra_tree
+  n_jobs: -1
+  verbose: 1
+  early_stop: true
+
+pycaret:
+  session_id: 42
+  n_select: 5
+  fold: 5
+  verbose: false
+  optimize: Accuracy
+  use_gpu: false
+  tuning:
+    enabled: true
+    n_iter: 10
+    optimize: Accuracy
+  ensemble:
+    enabled: false
+    method: Bagging
+    n_estimators: 10
+
+common:
+  target_column: target
+  problem_type: auto
+  cv_folds: 5
+  random_state: 42

config/config.yaml

@@ -0,0 +1,53 @@
+artifacts_root: artifacts
+
+data_ingestion:
+  root_dir: artifacts/data_ingestion
+  source_url: null
+  local_data_file: artifacts/data_ingestion/data.csv
+  unzip_dir: artifacts/data_ingestion
+
+data_validation:
+  root_dir: artifacts/data_validation
+  data_dir: artifacts/data_ingestion/data.csv
+  status_file: artifacts/data_validation/status.txt
+  schema_file: config/schema.yaml
+
+data_transformation:
+  root_dir: artifacts/data_transformation
+  data_path: artifacts/data_ingestion/data.csv
+  train_path: artifacts/data_transformation/train.csv
+  test_path: artifacts/data_transformation/test.csv
+  test_size: 0.2
+  random_state: 42
+
+feature_engineering:
+  root_dir: artifacts/feature_engineering
+  train_path: artifacts/data_transformation/train.csv
+  test_path: artifacts/data_transformation/test.csv
+  output_train_path: artifacts/feature_engineering/train_features.csv
+  output_test_path: artifacts/feature_engineering/test_features.csv
+
+model_trainer:
+  root_dir: artifacts/model_trainer
+  train_data_path: artifacts/feature_engineering/train_features.csv
+  test_data_path: artifacts/feature_engineering/test_features.csv
+  model_path: artifacts/model_trainer/model
+  target_column: target
+
+model_evaluation:
+  root_dir: artifacts/model_evaluation
+  model_path: artifacts/model_trainer/model
+  test_data_path: artifacts/feature_engineering/test_features.csv
+  metrics_file: artifacts/model_evaluation/metrics.json
+  target_column: target
+
+model_pusher:
+  root_dir: artifacts/model_pusher
+  model_path: artifacts/model_trainer/model
+  model_registry_path: models/production
+
+mlflow:
+  tracking_uri: http://localhost:5000
+  experiment_name: automl_experiment
+  run_name: null
+  registry_uri: null

config/deployment_config.yaml

@@ -0,0 +1,42 @@
+api:
+  host: 0.0.0.0
+  port: 8000
+  workers: 4
+  reload: false
+  log_level: info
+  title: AutoML MLOps API
+  version: 1.0.0
+
+model:
+  path: models/production
+  cache_enabled: true
+  cache_ttl: 3600
+  prediction_timeout: 30
+
+docker:
+  image_name: automl-mlops
+  registry: ghcr.io
+  tag: latest
+  platform: linux/amd64
+
+huggingface:
+  space_name: null
+  space_sdk: docker
+  space_hardware: cpu-basic
+  private: false
+
+health:
+  liveness_path: /health
+  readiness_path: /readiness
+  startup_timeout: 60
+
+metrics:
+  enabled: true
+  path: /metrics
+  include_trace_exemplar: true
+
+logging:
+  level: INFO
+  format: json
+  rotation: 100 MB
+  retention: 7 days

config/monitoring_config.yaml

@@ -0,0 +1,44 @@
+data_drift:
+  reference_data_path: artifacts/data/processed/reference.csv
+  current_data_path: artifacts/data/processed/current.csv
+  report_path: artifacts/reports/drift_reports
+  threshold: 0.1
+  columns: null
+
+model_monitoring:
+  predictions_path: artifacts/logs/predictions
+  performance_threshold: 0.05
+  window_size: 1000
+  metrics:
+    - accuracy
+    - precision
+    - recall
+    - f1
+
+evidently:
+  reference_window: 7
+  detection_window: 1
+  confidence_level: 0.95
+  stattest: ks
+  stattest_threshold: 0.05
+
+prometheus:
+  host: localhost
+  port: 9090
+  scrape_interval: 15s
+  evaluation_interval: 15s
+
+grafana:
+  host: localhost
+  port: 3000
+  admin_user: admin
+  admin_password: admin
+
+loki:
+  host: localhost
+  port: 3100
+
+promtail:
+  host: localhost
+  port: 9080
+  log_path: artifacts/logs

monitoring/dashboards/evidently_dashboard.ipynb

@@ -0,0 +1,10 @@
+{
+ "cells": [],
+ "metadata": {
+  "language_info": {
+   "name": "python"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

notebooks/01_data_exploration.ipynb

@@ -0,0 +1,10 @@
+{
+ "cells": [],
+ "metadata": {
+  "language_info": {
+   "name": "python"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

notebooks/02_automl_experiments.ipynb

@@ -0,0 +1,10 @@
+{
+ "cells": [],
+ "metadata": {
+  "language_info": {
+   "name": "python"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

notebooks/03_model_analysis.ipynb

@@ -0,0 +1,10 @@
+{
+ "cells": [],
+ "metadata": {
+  "language_info": {
+   "name": "python"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 5
+}

requirements-dev.txt

@@ -0,0 +1,19 @@
+-r requirements.txt
+
+pytest
+pytest-asyncio
+pytest-cov
+pytest-mock
+
+black
+flake8
+isort
+mypy
+
+pre-commit
+
+jupyter
+ipykernel
+notebook
+
+locust

requirements.txt

@@ -0,0 +1,30 @@
+fastapi
+uvicorn[standard]
+pydantic
+pydantic-settings
+
+pandas
+numpy
+scikit-learn
+
+autogluon.tabular
+flaml
+pycaret
+
+mlflow
+dvc
+
+evidently
+
+pyarrow
+fastparquet
+
+pyyaml
+python-box
+ensure
+python-multipart
+prometheus-client
+python-json-logger
+
+httpx
+requests