This repository contains the notebooks, Python scripts, and learning materials completed as part of the Machine Learning for Production (MLOps) Specialization, offered by DeepLearning.AI through Coursera.
The specialization focuses on the practical skills needed to design, build, deploy, and maintain machine learning systems in production. Unlike traditional machine learning courses that mainly focus on model training, this specialization covers the full production lifecycle: project scoping, data pipelines, data validation, modeling strategies, deployment patterns, monitoring, and continuous improvement.
The main goal of this repository is to organize the hands-on work completed throughout the specialization and document the most important concepts, tools, and workflows covered in each course.
Note: Datasets are not included in this repository for security, size, and distribution reasons. The repository is intended to store the course notebooks, Python scripts, and PDF guides organized by course and week.
The Machine Learning for Production (MLOps) Specialization is focused on taking machine learning projects beyond experimentation and preparing them for real-world use.
The specialization covers how to design production ML systems, define data requirements, create reliable data pipelines, validate and transform datasets, build modeling pipelines, optimize models for different serving environments, deploy models, and monitor them over time.
A key idea throughout the specialization is that production machine learning is not only about achieving good model performance. It also requires reliable data management, reproducible workflows, scalable infrastructure, deployment strategies, monitoring, and continuous improvement after the model is released.
The repository is organized by course and week. Each course has its own folder and README file. Inside each week folder, the corresponding notebooks, Python scripts, and PDF guides are included when available.
.
├── course_1_machine_learning_in_production/
├── course_2_machine_learning_data_lifecycle_in_production/
├── course_3_machine_learning_modeling_pipelines_in_production/
└── course_4_deploying_machine_learning_models_in_production/
Each course README provides a more specific summary of the weekly content, including the main techniques, tools, and libraries used.
Course 1: Machine Learning in Production
This course introduces the production machine learning lifecycle from an end-to-end perspective. It covers how to scope ML projects, define data and modeling requirements, choose deployment strategies, and identify common production challenges.
The course also focuses on practical issues such as setting a baseline, analyzing errors, handling concept drift, and building a framework for continuously improving a deployed ML system.
Course 2: Machine Learning Data Lifecycle in Production
This course focuses on the data side of production machine learning. It covers how to collect, clean, validate, transform, and manage datasets used in ML systems.
The course introduces important concepts such as data quality, feature engineering, feature selection, data lineage, provenance metadata, schemas, data evolution, labeling workflows, data augmentation, and preprocessing for different data types.
Course 3: Machine Learning Modeling Pipelines in Production
This course focuses on building and managing modeling pipelines for production environments. It covers how to design models for different serving requirements, manage modeling resources, optimize performance, and analyze model behavior.
The course also introduces topics such as neural architecture search, model resource management, high-performance modeling, model analysis, fairness, bottleneck detection, and interpretability.
Course 4: Deploying Machine Learning Models in Production
This course focuses on making machine learning models available to end users through reliable deployment systems. It covers model serving, infrastructure choices, real-time and batch inference, workflow automation, progressive delivery, monitoring, and logging.
The course also emphasizes how to keep production ML systems operating over time by detecting model decay, monitoring performance, and responding to system or model failures.
End-to-end ML system design Project scoping for production ML Data requirements and data definition Baseline modeling and error analysis Concept drift Data collection, cleaning, and validation Feature engineering and transformation Data lineage and provenance Data labeling and augmentation Modeling pipelines Neural architecture search Model resource management Model analysis and interpretability Model serving patterns Batch and real-time inference Progressive delivery Model monitoring and logging Continuous improvement of production ML systems
Specialization certificate: https://www.coursera.org/account/accomplishments/specialization/certificate/FP6RBW3ZMAUX
This specialization was offered by DeepLearning.AI through Coursera.
Course notebooks, scripts, slides, and educational materials are credited to DeepLearning.AI. Materials are included only for educational and non-commercial purposes, following the license terms provided in the course materials.