DeepLearning.AI Professional Certificates

Learning plan

About 80 hours of focused study, split into 4 phases. Pick your weekly pace and the timeline adjusts.

Estimated finish: 8 weeks

What the exam covers

Source: official exam guide. Weights can change, so check before you start.

Before you start

• Basic Python and high-school maths (functions, graphs, some algebra). The Machine Learning Specialization explains the calculus it needs as it goes
• Graded labs and certificates need a paid plan (Coursera subscription or DeepLearning.AI Pro). Short courses on deeplearning.ai are free to audit

1. Weeks 1–3 · 25 h

   Regression and classification (course 1)

   • Implement linear regression with gradient descent in NumPy, including vectorisation, feature scaling and learning-rate tuning
   • Implement logistic regression with a sigmoid, cross-entropy loss and decision boundaries
   • Diagnose overfitting and fix it with L2 regularisation

   Hands-on: Without scikit-learn, write gradient descent for linear regression on a small housing CSV. Plot cost against iterations for three learning rates and explain why one of them diverges.

   Use: Supervised Machine Learning: Regression and Classification · Machine Learning Specialization overview

2. Weeks 3–6 · 27 h

   Neural networks and decision trees (course 2)

   • Build multi-layer networks in TensorFlow, choose activations (ReLU, sigmoid, softmax) and train multiclass classifiers
   • Use bias/variance analysis, learning curves and error analysis to decide the next step on a model
   • Train decision trees, random forests and XGBoost, and know when trees beat neural networks on tabular data

   Hands-on: Train a Keras network on handwritten digits (MNIST) and an XGBoost model on a Kaggle tabular dataset. Compare their train and validation curves and write up the bias/variance diagnosis for each.

   Use: Advanced Learning Algorithms · TensorFlow tutorials

3. Weeks 6–8 · 22 h

   Unsupervised learning, recommenders and RL (course 3)

   • Implement k-means and Gaussian anomaly detection, and choose k and the threshold sensibly
   • Build collaborative-filtering and content-based recommenders
   • Explain the Bellman equation and deep Q-learning, and train an agent on the Lunar Lander lab

   Use: Unsupervised Learning, Recommenders, Reinforcement Learning

4. Week 8 · 6 h

   Bridge to applied GenAI (free short courses)

   • Complete two free DeepLearning.AI short courses on prompting and LLM application building
   • Pick your next certificate path (Deep Learning Specialization or a GenAI course) based on the gaps you found

   Use: ChatGPT Prompt Engineering for Developers · DeepLearning.AI short courses catalogue · Deep Learning Specialization (next step)

You're ready to book when…

• Every graded lab in all three courses is passed, so the specialization certificate is issued
• You can derive the gradient-descent update for logistic regression on paper
• You can explain a bias/variance diagnosis and the fix you would try for each case

Exam tips

• Coursera lists the ML Specialization at about 95 hours. If you already know Python and NumPy, the optional labs move fast and your total can be well below that.