MIT Professional Certificate in Machine Learning & AI

Learning plan

About 130 hours of focused study, split into 4 phases. Pick your weekly pace and the timeline adjusts.

Estimated finish: 13 weeks

What the exam covers

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

Before you start

• Budget and time. There's a $325 application fee, courses cost about $2,500–$4,900 each, and you need 16 course-days within 36 months, mostly on MIT's campus in June–August
• Linear algebra, probability and Python. Without prior ML experience you must take the Foundations course before Advanced

1. Weeks 1–3 · 25 h

   Pre-course maths and ML refresh

   • Refresh matrix operations, eigenvectors, gradients and probability (Bayes' rule, expectation, Gaussians)
   • Work through linear classifiers, the perceptron, logistic regression and regularisation in Python before you arrive

   Hands-on: Implement the perceptron and regularised logistic regression from scratch in NumPy on a text dataset (for example spam vs ham) using bag-of-words features. These are the models the core course builds on.

   Use: MIT OCW 6.036 Introduction to Machine Learning · MIT OCW 18.06 Linear Algebra

2. Weeks 3–7 · 40 h

   Core courses on ML for big data and text

   • Cover the core models used for text and large datasets, including linear classifiers, kernels, neural networks, clustering and recommender basics
   • In the Advanced course, apply deep learning, generative models and unsupervised methods to large-scale and text problems
   • Leave with a list of three ML problems from your own organisation scoped against what you learned

   Use: Core: ML for Big Data and Text Processing – Foundations · Core: ML for Big Data and Text Processing – Advanced

3. Weeks 7–11 · 45 h

   LLM and systems electives

   • Pick electives that add up to the 11 remaining days. One example path is AI System Architecture and LLM Applications (5 days), Ethics and Risks of AI (4 days) and Designing Efficient Deep Learning Systems (2 days)
   • Design an LLM application architecture (RAG, fine-tuning, evaluation, serving cost) for a real use case
   • Document risks (bias, privacy, misuse) and mitigations for that design

   Hands-on: Turn one of your three scoped problems into a working prototype, for example a RAG assistant over internal docs or a text classifier. Write a two-page design doc covering architecture, evaluation metrics, cost and risk mitigations, ready to present to your employer.

   Use: MIT Professional Education certificate program and elective list

4. Weeks 12–13 · 20 h

   Consolidate and apply at work

   • Present your prototype and design doc to your team or sponsor, with a go/no-go recommendation
   • Check with MIT Professional Education that all 16 days are recorded within the 36-month window

   Use: MIT OCW 6.867 Machine Learning (graduate-level depth)

You're ready to book when…

• You have completed at least one core course and 16 qualifying course-days in total
• You can explain the models from the core course and when you'd pick each for a text problem
• You have a prototype and design doc that applies the program to a real problem at your organisation

Exam tips

• Courses mostly run in June, July and August and fill up. Plan your 36-month window around two or three summers and apply early.
• If you have prior ML experience you may skip Foundations and start with Advanced, but you then need more elective days to reach 16.