Learn, build and prepare
Learn the model tooling, build a retrieval application, then add agent workflows where they solve a real problem. Evaluation and a reproducible project matter more than the number of courses completed.
Reviewed: · Published by BonusMantra
See the learning orderFollow the roadmap, choose relevant learning resources, build a project, then test your understanding with interview practice.
The sequence, role fit and project exercises are editorial recommendations. Provider requirements and certificate conditions are shown separately.
These resources offer free learning access. Some are guided tutorials rather than full courses. A free course does not necessarily include a certificate, hosted compute, private-repository features or a professional exam.
1. Hugging Face · Course
Useful foundations for understanding the components around a retrieval system and the limits of a model pipeline.
Inspect how text is tokenized and compare model outputs for differently sized inputs.
Source checked:
2. DataTalks.Club · Course
A practical fit for learners who want to build and evaluate a knowledge-base assistant.
Create a small question set, compare retrieval results and record which answers lack supporting evidence.
Source checked:
3. Hugging Face · Course
Useful for moving beyond a single model call and understanding tool use, agent frameworks and an agentic RAG example.
Add one well-scoped tool to a sample agent and log when it is called. Explain when a simpler retrieval flow would be enough.
Source checked:
Pick one resource, complete a small exercise and explain the result aloud before adding another course.
