AI-ML-Roadmap-from-scratch: A Curated Link Index, Not a Course
Become skilled in Artificial Intelligence, Machine Learning, Generative AI, Deep Learning, Data Science, Natural Language Processing, Reinforcement Learning and more with this complete 0 to 100 repository.
At a glance
- What is it?
- Aadith Sukumar's repository is a module-ordered list of free AI and ML resources, MIT licensed, with no code of its own. It is useful as a syllabus skeleton and useless as a substitute for the courses it points to.
- Who is it for?
- Adopt this repository if you want a pre-sorted reading order for free AI and ML material and you are willing to check every link yourself before committing time to it. Do not adopt it if you need graded exercises, runnable notebooks or an offline copy, because the repository ships none of those.
- Can I use it commercially?
- Yes. MIT is a permissive licence: you can use, modify and sell software built on it, as long as you keep its copyright and licence notices.
- Is it still maintained?
- Yes. The repository last received commits 34 days ago.
- What is it written in?
- GitHub does not report a main language for this repository.
Answers come from the project's GitHub data, last synced on September 15, 2026, and from our analysis. They are not legal advice.
DEEP OPEN-SOURCE ANALYSIS
What the Repository Actually Contains
The README describes the project as a collection of free resources from YouTube, online courses, blogs and websites, ordered into modules that increase in difficulty. That is the whole artifact. There is no training code, no dataset, no notebook and no library published under this name. The primary language field is listed as unknown, which is consistent with a repository whose main body is Markdown tables.
The structure runs from Module 0 (Before You Start) through Module 10 (Agentic AI), with a Sub-Module 7A covering Retrieval Augmented Generation, and a Bonus Module for advanced pathway courses. Each module is a table with three columns: serial number, a type label such as Software, Py Package, Playlist, Course, Lectures, Website, YouTube, Certificate or Practice, and the course name as a hyperlink. A star emoji marks entries the author recommends most highly.
This matters for adoption decisions. You are not installing a framework. You are deciding whether someone else's link ordering saves you more time than it costs you to audit. For a self-directed learner facing an unbounded search space, a ranked list has real value. For an engineering team that wants reproducible training material, the repository offers nothing to run.
The Module Ordering and Its Implicit Prerequisites
The sequence is math first, then programming foundations, then data science, then classical machine learning, then computer vision, deep learning, generative AI, NLP, reinforcement learning and finally agentic AI. The README states the modules can be followed simultaneously as well as in order, which softens the sequence into a suggestion rather than a dependency chain.
That flexibility is also the weakest part of the design. Module 1 lists a Math for Machine Learning playlist, an NPTEL Discrete Mathematics course, a Saylor Academy Discrete Structures course, MIT's Linear Algebra lecture series and CodeCademy's Fundamental Math for Data Science. Five math tracks with no indication of which to pick if you only want one. The star system flags NPTEL Discrete Mathematics as recommended, but a learner aiming at deep learning would get more from the linear algebra series, and the README does not say so.
Module 2 repeats the same paragraph that opens Module 1, word for word, about the domain being a vast deep ocean and building a boat. That is a copy-paste artifact, and it is the kind of detail that tells you how much editorial review the prose received. The tables are the substance; the surrounding text is scaffolding.
Setting Up the Environment the Roadmap Assumes
Module 0 is the only module with concrete setup instructions, and they are links rather than commands. It points to the Python 3.14 download page, the Visual Studio Code download page, a GeeksforGeeks article on installing pip on Windows, and a file in the repository itself at Packages.md described as Common Python Libraries used for AI/ML.
The README does not give a pip install line, a virtual environment command, or a requirements file. If you want to follow the intended path, the sequence implied by the table is: install Python from python.org, install VS Code from code.visualstudio.com, then follow the linked pip installation guide, then open Packages.md for the library list. The Windows-specific framing of the pip link is worth noting, since the repository does not appear to link a macOS or Linux equivalent in Module 0.
Packages.md is the one file in the repository that carries project-specific content beyond link curation, and it is the first thing to read before trusting the rest. If that file is current, the repository has a working entry point. If it is stale, the entire setup module is a dead end.
Where the Curation Breaks Down
Link rot is the structural risk. Every module is a table of external URLs pointing at YouTube playlists, Coursera professional certificates, edX courses, NPTEL Swayam pages, MIT OpenCourseWare downloads and personal blogs. None of those are under the author's control. A YouTube playlist can be made private, a Coursera certificate can be restructured, an NPTEL course page can move between semesters. The repository has no mechanism described for detecting or repairing broken links.
There is also no difficulty calibration beyond the star markers and the stated increasing order. A learner cannot tell from the table whether the IBM Data Science Professional Certificate in Module 3 takes forty hours or four hundred, or whether the Google Data Analytics certificate overlaps with it. Two professional certificates listed in the same module with no guidance on choosing between them is a decision the reader has to make alone.
Finally, the repository assumes you will complete courses on third-party platforms. Progress tracking, certificates and exercises live there, not here. If a platform changes its free tier, the roadmap's promise of free resources weakens without the repository changing a line.
What a Maintained Curriculum Repository Looks Like Instead
The closest comparison is a structured curriculum project such as Microsoft's ML for Beginners, which ships Jupyter notebooks, datasets and lesson text inside the repository itself. The difference in approach is fundamental. ML for Beginners gives you material you can run offline and version alongside your own work; AI-ML-Roadmap-from-scratch gives you pointers to material hosted elsewhere.
That is not automatically worse. A notebook-based curriculum locks you into its authors' choice of framework, dataset and pace, and it goes stale the moment a library API changes. A link index stays valid longer per unit of maintenance, because a link to a linear algebra lecture series does not break when PyTorch changes its optimizer signature. The trade-off is that you get no executable artifact and no guarantee the linked course still teaches what the table implies it teaches.
For a learner who already knows how to work through video courses and wants a syllabus, the link index wins on breadth. For anyone who needs to run code today, it loses outright.
Licence, Contributions and Maintenance Signals
The repository is MIT licensed. For a collection of links and Markdown tables, the practical effect is that you can copy the module structure into your own notes, fork it for a study group, or republish the ordering with attribution, without asking. MIT does not grant rights to the linked third-party content, and the courses, videos and certificates the tables point to carry their own terms. Copying the roadmap does not copy the courses.
The topics list includes hacktoberfest and hacktoberfest2025, and the README has a Contribute section, so the project is positioned to accept pull requests, typically link additions. The last push timestamp is 2026-08-12, which indicates recent activity, though no releases have been published and none would be expected for a document-only repository.
Maintenance cost for a user is low and recurring: periodically re-check the links in the modules you actually intend to follow. There is no CI job described that would do this for you, and the repository does not document a link-checking workflow.
Who Should Start at Module 0 and Who Should Close the Tab
Start here if you are self-taught, budget-constrained and unsure what order to learn things in. The module sequence from math through to agentic AI covers a wider surface than most single-platform curricula, and the star markers give you a first pass at triage. The two YouTube walkthrough videos linked at the top of the README let you preview the author's reasoning before committing to the tables.
Close the tab if you learn by building, if you need a graded path with feedback, or if your employer requires completion certificates tied to a specific platform. Also close it if you are looking for a library to depend on. There is nothing here to import.
Before you begin, verify three things in this order: that Packages.md still lists libraries you can install with your current Python version, that the Python 3.14 download page and the pip installation guide both load, and that the specific course you intend to start with is still free and still available in your region. Those three checks take minutes and determine whether the rest of the roadmap is usable. If Packages.md is out of date, treat Module 0 as the only part of the repository you need to fix yourself before proceeding.
Editorial conclusion
Adopt this repository if you want a pre-sorted reading order for free AI and ML material and you are willing to check every link yourself before committing time to it. Do not adopt it if you need graded exercises, runnable notebooks or an offline copy, because the repository ships none of those. Before starting Module 1, open the Packages.md file and the Python download page in the same browser session and confirm both resolve, since a broken toolchain link at Module 0 stalls everything downstream.
Community notes