Machine Learning from Scratch: The Systems View (Advanced)
A 6-week intensive program designed to transform ML practitioners into ML Systems Engineers by building a complete mini-PyTorch framework from first principles. Learners will implement automatic differentiation, neural network layers, optimizers, and GPU acceleration basics while understanding the computational and numerical considerations that drive production ML systems. By coupling each implementation with performance optimization and framework comparisons, students develop the deep systems knowledge required for mid-level engineering roles where architectural decisions and production trade-offs are daily responsibilities.
6
Modules — Structured path
46
Interactive AI lessons
Flexible
Self-paced
Intermediate
Recommended experience
What You'll Learn
Implement a complete automatic differentiation engine supporting both scalar and tensor operations with dynamic computational graph construction
Build neural network primitives (linear layers, activations, batch normalization) from scratch with proper initialization strategies to prevent vanishing/exploding gradients
Code and analyze gradient-based optimization algorithms (SGD, Momentum, Adam) understanding their convergence properties and hyperparameter sensitivities
Write and profile basic CUDA kernels to understand GPU memory hierarchies and parallelism trade-offs in ML systems
Assemble a trainable mini-PyTorch framework and train a convolutional neural network achieving >70% accuracy on CIFAR-10
Benchmark custom implementations against PyTorch to understand the engineering trade-offs in production ML systems
Skills You'll Gain
124 skillsProgram Structure
Meet Your Expert
Firas Bayram
Machine LearningArtificial Intelligence researcher and university instructor specializing in adaptive learning systems, large language models, and generative AI. Passionate about bridging cutting-edge AI research with real-world applications and helping leaders adopt modern practical, data-driven AI solutions.
Ready to start learning?
Sign up for free and enroll in this program.