Course Objective & Scope
Many enterprise production pipelines, mobile/embedded deployments and Google Cloud (Vertex AI, TFX) environments standardise on TensorFlow and Keras. This course covers flexible network design with the Keras 3 Functional and Model Subclassing APIs, high-throughput data pipelines using tf.data and TFRecords, compiled custom training steps with tf.GradientTape and @tf.function, and preprocessing layers that eliminate training-serving skew.
Multi-GPU strategies with tf.distribute, TensorBoard profiling and exporting models to SavedModel, TensorFlow Serving and LiteRT (TFLite) are practised hands-on. Participants produce a production-ready TensorFlow model artifact with in-graph preprocessing tailored to an enterprise dataset.
Who Should Attend?
- Data scientists and ML engineers building on TensorFlow/Keras enterprise stacks
- Teams deploying models to Vertex AI, TF Serving or mobile/edge targets (TFLite)
- Developers looking to customise and accelerate existing Keras workflows
- Prerequisite: Python and basic machine learning knowledge
Key Takeaways
- Build multi-input, multi-output and custom-layer architectures with the Keras Functional and Subclassing APIs
- Engineer high-throughput input pipelines with tf.data prefetching, interleaving and TFRecords
- Write fast graph-compiled custom training steps using tf.GradientTape and @tf.function
- Embed preprocessing layers inside the model graph to prevent training-serving skew
- Export and validate SavedModel artefacts for TensorFlow Serving and LiteRT (TFLite)
Curriculum
TensorFlow 2, Eager/Graph Execution and Keras 3
- tf.Tensor, tf.Variable and AutoGraph compilation with @tf.function
- Sequential, Functional API and Model/Layer Subclassing
tf.data Pipelines and Preprocessing Layers
- TFRecords, parallel reads, caching and prefetch optimisation
- In-graph preprocessing with Normalization, StringLookup and TextVectorization
Custom Training Loops and Callbacks
- tf.GradientTape, custom losses, metrics and overriding train_step
- Hyperparameter tuning with KerasTuner and TensorBoard profiling
Distributed Training and Performance
- Scaling with MirroredStrategy and MultiWorkerMirroredStrategy
- Mixed precision policies and XLA compiler acceleration
SavedModel, TF Serving and TFLite Deployment
- Defining serving signatures and exporting SavedModel artefacts
- Serving via REST/gRPC in TF Serving and LiteRT (TFLite) conversion
Other Courses in This Category
-
Quantisation (PTQ, QAT, INT8/INT4), structured pruning, knowledge distillation and deployment with ONNX Runtime, TensorRT, OpenVINO and LiteRT (TFLite) for lower cloud GPU costs and fast edge inference.
- Live Course
- 5 modules
-
Foundations and architectures of deep generative modelling: VAEs, GANs, Normalizing Flows, Denoising Diffusion (DDPM/Latent Diffusion), Flow Matching and autoregressive models for synthetic data and multimodal generation.
- Live Course
- 5 modules
-
MDPs, Q-Learning, DQN, Policy Gradients (PPO, SAC), contextual bandits and Offline RL using Gymnasium and Stable-Baselines3 for dynamic pricing, resource allocation and sequential control problems.
- Live Course
- 5 modules
-
Deep fraud models over transaction sequences, session clickstreams and payment networks: Transformer/RNN sequence encoders, autoencoder anomaly scores, GNNs for fraud rings and hybrid GBDT+DL architectures.
- Live Course
- 5 modules
-
Deep defect and fault detection across inspection images, acoustic/vibration signals and system logs using PatchCore/PaDiM, Anomalib, YOLO/segmentation, 1D-CNNs/autoencoders and normal-only training.
- Live Course
- 5 modules
-
Multivariate, long-horizon and probabilistic time-series forecasting with N-BEATS, N-HiTS, Temporal Fusion Transformer (TFT), PatchTST, DeepAR and TCN using NeuralForecast and PyTorch Forecasting.
- Live Course
- 5 modules
Get in touch
