AI / ML Learning Roadmaps: A Structured Resource Hub from Beginner to Research

A Navigation System for the AI/ML Learning Ecosystem

The core problem this repository addresses is that the AI and machine learning field has more high-quality free resources than any single learner can sensibly evaluate. Course platforms, textbooks, YouTube channels, research paper reading lists, and MLOps guides all exist independently, and a beginner starting from scratch has no obvious way to judge which to prioritise or in what order.

bishwaghimire/ai-learning-roadmaps is a curated index that makes those decisions explicit. It points to external resources and organises them by topic, difficulty, and career goal. The audience is described in the README as computer science students, AI enthusiasts, and working professionals who want to build strong foundations and progress confidently from beginner to advanced levels. That is a wide range, and the repository handles it by splitting the material into modular roadmaps rather than a single linear track.

It is not a course and does not distribute any learning content itself. The repository holds Markdown roadmap files and reference tables. A learner who opens it will find structured lists of links, not exercises or videos.

Repository Layout and Roadmap Coverage

The top-level directory contains README.md, LICENSE, CONTRIBUTING.md, Packages.md, and three subdirectories: roadmaps/, resources/, and assets/. The roadmaps/ directory is where all the structured content lives. Each roadmap is a separate Markdown file covering a distinct topic:

- roadmaps/ai-roadmap.md covers big-picture AI concepts, history, and paradigms. - roadmaps/ml-roadmap.md covers supervised, unsupervised, and classical machine learning. - roadmaps/deep-learning-roadmap.md covers neural networks, CNNs, RNNs, and Transformers. - roadmaps/nlp-roadmap.md covers text processing, Transformers, and modern NLP systems. - roadmaps/llm-roadmap.md covers pretraining, fine-tuning, alignment, and evaluation. - roadmaps/rag-roadmap.md covers vector search, embeddings, and system design. - roadmaps/mlops-production-ai-roadmap.md covers deployment, monitoring, and scalability. - roadmaps/research-scientist-roadmap.md is aimed at PhD-level research. - roadmaps/ai-safety-alignment-roadmap.md covers ethics, governance, and alignment.

The Packages.md file lists essential Python libraries for AI and ML work. The math resources section of the README provides a subject-by-subject table that includes specific resources such as the 3Blue1Brown Essence of Linear Algebra playlist and the MIT OCW Linear Algebra 18.06 course, tagged by difficulty level and type.

Cloning the Repository and Starting Your Path

There is no installation beyond cloning the repository and reading its files. No build step, no Python environment, and no server are required.

git clone https://github.com/bishwaghimire/ai-learning-roadmaps.git
cd ai-learning-roadmaps

Once inside, the README suggests a specific entry point for new learners: start with the AI Roadmap if you are new to the field, then move into ML, then deep learning, then a specialisation such as computer vision, NLP, or LLMs. For learners who already have some background, the career track table in the README maps each role to an ordered sequence of roadmap files.

The Packages.md file lists the Python libraries you will need once you begin working through the external courses and notebooks. The README also lists five tool prerequisites for setting up a development environment: Python 3.10 or later, VS Code, Python venv for environment isolation, and either Google Colab or a local Jupyter Notebook for running code. These tools are not bundled in the repository; the README points to their official download pages.

Because all roadmap content is plain Markdown, you can read everything offline after cloning.

Career Tracks: How the Roadmaps Connect

The most practically useful part of the README is the career-oriented learning paths table. It maps ten distinct roles to ordered sequences of roadmap files, which removes the most common source of confusion for learners who are unsure where to start:

- AI Engineer: AI Roadmap, then ML, then deep learning, then CV or NLP, then LLMs. - Data Scientist: Math resources, then Python, then ML, then Statistics, then Data Science Roadmap. - GenAI Engineer: AI, then deep learning, then LLMs, then Generative AI, then RAG. - MLOps Engineer: ML, then deep learning, then the MLOps and Production AI Roadmap. - Research Scientist (PhD-level): ML, then deep learning, then theory, then the Research Scientist Roadmap. - AI Safety and Policy: AI Roadmap, then LLMs, then the AI Safety and Alignment Roadmap.

The README states that these sequences are guidelines, not strict rules, and that the roadmaps are modular but connected. A learner can follow them sequentially or start from any roadmap that matches their existing background. The README explicitly says you do not need to follow everything linearly.

This modularity is a real design choice. Someone who already knows Python and basic ML can go directly to the LLM Roadmap without reading the ML Roadmap from scratch.

What This Repository Does Not Provide

The repository is a curated index, not a self-contained learning system. Several things a learner might expect are not here.

There are no exercises. The roadmaps link to courses and textbooks that have exercises, but the repository itself contains no problem sets, code notebooks, or autograded tasks. A learner who needs structured practice with immediate feedback will have to set that up through the external resources the roadmaps point to.

There is no community built into the repository. The repository has a CONTRIBUTING.md and accepts pull requests, but it does not include a forum, a Discord server, or any synchronous support structure. This is not a criticism; it is a scope boundary worth knowing before starting.

The README also does not guarantee that all linked external resources are free. Most of the listed courses and playlists are publicly accessible, but some linked textbooks are commercial. Learners should check each resource's access model before building a study plan.

Finally, the repository does not explain concepts itself. Each roadmap file lists what to study and in what order, but does not teach the material. It is a map, not the destination.

A Comparable Alternative and the Key Difference

Roadmap.sh is a browser-based platform that provides interactive, visually navigable skill roadmaps for software engineering roles: frontend, backend, DevOps, and others. It is widely used and maintained as a web application with a visual graph interface that lets users mark their progress. The key difference is scope: Roadmap.sh targets software engineering broadly, with AI/ML coverage as one section among many, while bishwaghimire/ai-learning-roadmaps focuses entirely on AI, ML, deep learning, LLMs, and research, with considerably more depth in those areas. Roadmap.sh includes interactive checkboxes and community-contributed paths; this repository is a plain Markdown collection that you read locally.

For learners whose primary goal is AI and machine learning with research-grade coverage including AI safety and alignment, the more focused structure here will be more useful. For learners who want a visual progress tracker or who are studying a mix of AI and general software engineering, Roadmap.sh covers that wider surface.

The last push to bishwaghimire/ai-learning-roadmaps was on 2026-06-15. The repository is MIT-licensed.

Editorial conclusion

bishwaghimire/ai-learning-roadmaps is a practical starting point for self-directed AI learners who want a structured reading list with career-track guidance. It is not suitable for anyone who needs structured exercises, autograded assignments, or a community forum built into the learning environment. Before committing to a career track, confirm that the roadmaps directory contains the specific file for that track (for example, roadmaps/rag-roadmap.md for RAG, or roadmaps/ai-safety-alignment-roadmap.md for AI safety) and verify that the referenced external links are still active, since the repository does not host the linked courses or books itself.

Frequently asked questions

Official sources

1. bishwaghimire/ai-learning-roadmaps on GitHub
2. Issues
3. License: MIT
4. README