TensorFlow in Action

Hands-On Foundations for TensorFlow 2 💻

Thushan Ganegedara’s TensorFlow in Action takes a deliberately applied approach to deep learning. Instead of stopping at API basics, it walks through building, training, improving, and shipping models with TensorFlow 2 and Keras.

The first part of the book establishes a solid foundation. You’ll see how TensorFlow 2 operates under the hood, work with tensors, variables, and operations, and explore Keras model-building APIs, including the Sequential API, functional API, and sub-classing API. Chapters on data retrieval introduce, Keras DataGenerators, and the tensorflow-datasets package—skills that quickly become essential when managing real-world data pipelines.

Building Real-World Deep Networks 🧠

From there, the book moves into architectures you’ll actually use. You’ll implement fully connected networks, convolutional neural networks, and recurrent neural networks, then tackle Transformers—the architecture that has reshaped natural language processing.

Later chapters go deeper into practical computer vision and NLP. Image classification with Inception-style networks, improving CNNs with regularization and transfer learning, Grad-CAM explainability, and image segmentation all receive detailed treatment. On the language side, you’ll build sentiment analysis and language modeling systems, explore sequence-to-sequence learning, and gain a practical understanding of Transformer components like self-attention and multi-head attention.

From Model Monitoring to Deployment ⚙️

Beyond model building, the book covers tools that matter in production. TensorBoard is presented as a way to monitor training, visualize metrics, and debug complex models. TFX chapters address MLOps and the process of deploying models with TensorFlow. This emphasis on observability and deployment makes the book useful for practitioners who need to move beyond notebooks.

Who Should Read This Book

Readers with some Python familiarity and a basic understanding of machine learning will benefit most from the book’s practical, example-driven style. Whether you are an engineer, data scientist, or developer moving into deep learning, the material provides a clear path from core TensorFlow concepts to production-oriented workflows.