NewThird edition in preparation, with diffusion models and large language models

Second edition · Real-world deep learning interview problems & solutions

Deep Learning
Interviews

Shlomo Kashani

A working reference for interviews and graduate-level exams

Hundreds of fully solved job interview questions from key AI topics.

400Pages

100+Problems

7Chapters

Free PDF Buy on Amazon GitHub ★ --

Deep Learning Interviews, second edition, by Shlomo Kashani — Second edition · View full-size cover

The book

Prepare to speak
fluently on AI.

An essential resource for aspiring data scientists and AI professionals facing competitive job interviews or graduate-level exams.

This comprehensive guide provides step-by-step solutions, clear diagrams, and detailed reasoning to make complex concepts intuitive.

Explore the seven chapters

Third edition · In preparation

The next edition
covers diffusion and LLMs.

The third edition joins this book with The Art and Science of Generative AI. It keeps the deep learning core and adds diffusion models, large language models, and the engineering that trains and serves them.

Interviewers now ask about forward and reverse diffusion, the KV cache, and how a team aligns and serves a large model. You work each answer out from first principles, line by line.

640Solved problems

21Chapters

124Figures

Mock cover of Deep Learning Interviews, third edition, by Shlomo Kashani — Third edition · Cover mock-up

Full solutions
Every problem now comes with a full solution, derived in numbered equations you can follow with pencil and paper.
Levels and roles
Each problem carries a level from 1 to 4 and a role, so you can work the core problems first when an interview is close.
Figures in one style
The figures share one drawing style and print in grayscale, so they read the same on paper as on a screen.

Four parts, 21 chapters

Mathematical Foundations

Logistic regression, Bayesian deep learning, information theory, deep learning calculus, and the probability behind generative models.

Deep Neural Networks

Cross validation and convolution, perceptrons and activation functions, metrics and training, ensembles, feature extraction and transfer learning, and the architectures of generative models.

III

Diffusion Models

Generative models and the diffusion idea, stochastic calculus, forward and reverse diffusion, and denoising diffusion in practice.

New in the third edition

Large Language Models

Autoregressive language models, tokens and scaling, efficient attention, training at scale, alignment, and the systems that cache, batch and serve a model.

New in the third edition

The third edition is not on sale yet, and its cover may change. The second edition stays available as a free PDF and in print.

Inside Volume I

A comprehensive curriculum covering the mathematical and computational foundations of deep learning.

Information Theory

Entropy, mutual information, KL divergence, and their applications in machine learning and neural networks.

Calculus & Algorithmic Differentiation

Chain rule, backpropagation, automatic differentiation, and computational graph optimization techniques.

Bayesian Deep Learning

Probabilistic programming, uncertainty quantification, variational inference, and Bayesian neural networks.

Logistic Regression

Maximum likelihood estimation, regularization, feature engineering, and classification fundamentals.

Ensemble Learning

Bagging, boosting, random forests, gradient boosting, and model combination strategies.

Feature Extraction

Dimensionality reduction, PCA, t-SNE, feature selection, and representation learning techniques.

Deep Learning

Comprehensive treatment covering neural networks, CNNs, RNNs, attention mechanisms, and modern architectures with PyTorch, Python, and C++ examples.

Extended chapter · 100+ pages

Preview the book

Browse through the comprehensive content and see the quality of explanations firsthand.