Wirewalk Instituteof AI
mod-203 Intermediate Waitlist

Classical Machine Learning in Production

For data scientists and engineers who need tabular models that are validated honestly, explained to stakeholders and kept healthy after deployment.

The curriculum is being prepared and reviewed. Enrollment is not open yet.

About this course

For forecasting demand, scoring risk, predicting churn or flagging fraud on structured data, gradient-boosted trees and well-regularized linear models are usually the right tool, and a large language model is usually not. What separates a model that works in production from one that only worked in a notebook is rarely the algorithm. It is leakage, the validation scheme, calibration, and what happens when the data shifts.

This course builds scikit-learn Pipelines and ColumnTransformers that cannot leak, chooses time-based, grouped or stratified cross-validation to match the problem, and compares XGBoost, LightGBM and CatBoost on the same task. You tune with Optuna, calibrate probabilities, pick thresholds using business costs, and explain predictions with SHAP. The production half covers MLflow tracking and model registry, batch versus online scoring, ONNX export, drift and performance monitoring, and retraining policy. Library features are current as of October 2026.

What you will be able to do

  • Build scikit-learn Pipelines with ColumnTransformer that prevent train/test leakage
  • Choose a validation scheme (time-series, grouped, stratified) that matches how the model will be used
  • Train and compare gradient-boosted tree libraries and tune them with Optuna and early stopping
  • Calibrate predicted probabilities and set decision thresholds from explicit error costs
  • Explain global and individual predictions with SHAP and recognize where explanations mislead
  • Package, register and serve a model with MLflow in batch or online mode
  • Design drift and performance monitoring with a retraining trigger and rollback plan

Syllabus

6 modules · 19 lessons · about 14 hours

Module 1: Framing tabular problems

Turning a business question into a target, a metric and a baseline.
3 lessons
  1. When classical ML beats deep learning and LLMs
  2. Defining the target, the prediction time and the metric
  3. Baselines you must beat before anything else

Module 2: Pipelines and validation that do not lie

Feature engineering and evaluation set up so the score means something.
3 lessons
  1. scikit-learn Pipelines and ColumnTransformer
  2. Target leakage, temporal leakage and how to find them
  3. Cross-validation for time series, groups and imbalanced classes

Module 3: Models that win on tables

Regularized linear models and the three main gradient-boosting libraries.
4 lessons
  1. Regularized linear and logistic regression as strong baselines
  2. XGBoost, LightGBM and CatBoost compared
  3. Categorical features, missing values and class imbalance
  4. Hyperparameter tuning with Optuna and early stopping

Module 4: Trustworthy predictions

Making scores usable for decisions and explainable to the people who rely on them.
3 lessons
  1. Probability calibration: Platt scaling and isotonic regression
  2. Choosing thresholds from costs, capacity and precision-recall
  3. SHAP values, partial dependence and their limits

Module 5: Shipping the model

From a trained estimator to a versioned, served artifact.
3 lessons
  1. Experiment tracking and the model registry with MLflow
  2. Batch scoring versus online inference APIs
  3. Serialization pitfalls: pickle, skops and ONNX export

Module 6: Operating models over time

Knowing when a deployed model has gone stale and what to do about it.
3 lessons
  1. Data drift, concept drift and delayed labels
  2. Monitoring metrics, alerts and dashboards for tabular models
  3. Retraining policies, shadow deployments and rollback

Exam and certificate

The exam unlocks once you have completed every lesson. It draws 20 questions from a larger bank, spread across the modules, with the answer options shuffled for each attempt. You have 45 minutes and need 80% to pass. After a failed attempt you can try again after 24 hours, up to 3 attempts in total. Feedback is given per module, never as the answer key.

Passing earns a certificate of completion and assessment from the Wirewalk Institute of AI with a unique ID that anyone can check at institute.wirewalk.com/verify. It is not an accredited qualification or a vendor certification. How certification works.

The Wirewalk Institute of AI is independent. This course teaches publicly documented products and is not affiliated with or endorsed by scikit-learn, XGBoost, LightGBM, CatBoost, Optuna, SHAP, MLflow, or ONNX. Product and company names are trademarks of their respective owners.

More in Model Development

All 16 courses in this track
mod-201 Intermediate Available

Deep Learning with PyTorch

A hands-on course in modern PyTorch: tensors and autograd, nn.Module design, data loading, a training loop you understand line by line, GPU and mixed-precision training, torch.compile, and exporting a model for deployment.

14 h · 19 lessons See price at checkout
mod-202 Intermediate Available

Transformers from the Ground Up

Build a decoder-only transformer piece by piece: tokenization, embeddings, scaled dot-product attention, multi-head and grouped-query attention, positional schemes such as RoPE, and the training and sampling loop, then map it onto real open-weight model architectures.

16 h · 19 lessons See price at checkout
mod-204 Intermediate Waitlist

Computer Vision Models

From CNNs and vision transformers to self-supervised backbones, real-time detectors and promptable segmentation: how modern vision models work, how to fine-tune them on your own labeled data, how to measure them with mAP and IoU, and how to deploy them fast.

15 h · 18 lessons Join the waitlist