Image by Author
AI is reshaping the future of work.
The global AI market is expected to grow to more than $638 billion in 2028. Companies are actively investing in AI solutions to drive efficiency and business growth.
AI skills are no longer optional.
Existing roles like software development, product management, and data science now require some level of AI literacy — such as the ability to integrate LLMs into software applications, or the skill to combine analytics with generative AI to gain deeper insight into data.
Additionally, we are seeing a rise in the demand for specialized AI-related roles, such as LLM engineer, AI engineer, and AI architect.
Whether you want to stay on top of your field and get a higher salary or are simply looking for employment in tech, AI skills are a must.
And in this article, I have curated three courses that will equip you with the generative AI skills needed to stay on top of today’s tech market.
Course 1: Microsoft’s Generative AI for Beginners
Microsoft’s “Generative AI for Beginners” is one of the most comprehensive generative AI courses I’ve found online, and it’s completely free.
This program is structured as a GitHub repository and has over 20 lessons that will teach you both the theoretical and practical aspects of generative AI.
Here are some concepts that will be covered in this generative AI course:
1. Foundations behind generative AI models
This learning path starts from the very basics — explaining what generative AI is and how you can select the right model for specific tasks.
2. Prompt engineering
You will also learn how to prompt AI to give you your desired outcome. Techniques like zero-shot and few-shot prompting will be covered, and you will also learn how to limit an AI model’s output to be more predictable.
3. Text and image generation
After learning the theoretical AI concepts mentioned above, this course will teach you to build actual text and image generation models by implementing more complex technologies like semantic search.
These technologies are currently being implemented by large tech companies like Notion and Spotify, and learning them will make you more employable in the age of AI.
4. RAG Implementation
Retrieval Augmented Generation (RAG) is a technique used to combine LLMs with reliable knowledge bases to improve their performance at a specific task.
This is one of the most popular use cases for LLMs, and data scientists at tech companies (like mine) are currently working on building reliable RAGs using our company’s databases and documents.
This course will teach you to build RAG systems with vector databases and embedding techniques —in-demand skills that make you an attractive candidate in today’s tech market.
5. AI Agents
AI agents have created a new wave of hype in 2025. There are countless articles on how AI agents will transform the way we work by automating tasks and improving workplace efficiency.
In simple terms, an AI agent is a system that is autonomously able to plan and execute workflows without consistent human direction.
Microsoft’s generative AI course will teach you:
- What AI agents are
- The types of AI agents
- How to build an AI agent application to pitch a new product idea
6. Fine-Tuning Large Language Models
Fine-tuning is the process of tweaking the parameters of a language model using your own database.
This will help you customize a foundational model for your specific use case, which increases the performance of the model for your task.
Microsoft’s AI program will walk you through the process of fine-tuning Open AI’s GPT-3.5 models for a specific domain. You will also learn how to deploy and use the fine-tuned algorithm.
Course 2: Hugging Face Reasoning Course
The Hugging Face Reasoning Course is part of the larger Hugging Face LLM course, with a focus on building reasoning models.
This course follows a cohort-based structure, and different units are released progressively at different times, making it the perfect course for anyone who enjoys learning with a structured timeline.
I recommend taking this course if you already have some knowledge of LLMs and would like to understand how to improve a language model’s reasoning capabilities.
Here are the topics that will be covered in the above course:
1. Fundamentals of Reasoning in LLMs: You will get an introduction to reinforcement learning concepts, and how they can be used to improve a language model’s reasoning capabilities.
2. Understanding the DeepSeek R1 Paper: You will learn about the research behind DeepSeek R1 — specifically how the model is able to learn through trial and error using reinforcement learning.
3. Advanced Interpretation of GRPO: GRPO is a reinforcement technique used in models like DeepSeek R1, that reduces the computational requirements compared to traditional techniques used in models like ChatGPT.
You will learn about this technique and understand how models can be fine-tuned with minimal hardware.
GRPO with Unsloth: You will learn to fine-tune models with GRPO using the Unsloth library.
In addition to the above topics, this course also has an interactive code review session, and live sessions on building Open R1 (a completely open reproduction of the Deepseek R1 model).
Note: This is a live course, and lessons are progressively released every week. You can find materials for previous lessons in this link.
Course 3: Hugging Face Agents Course
The Hugging Face Agents Course teaches you how to build and deploy AI agents. As explained earlier in this article, AI agents are systems that can autonomously take action to achieve specific goals.
Here are the topics you will learn in Hugging Face’s Agents Course:
1. Intro to AI Agents: You will learn what AI agents are, and get a recap of how LLMs work. You will also learn to create your own AI agent in this lesson.
Frameworks for AI Agents: This lesson will cover three popular AI agent frameworks that will simplify the process of building and deploying complex workflows.
2. Use Case for Agentic RAG: RAGs, as explained previously in this article, can help you retrieve relevant information from your own database and forward it to LLMs. In this lesson, you will learn about agentic RAGs, which will use agents to answer questions about your data.
3.Final Project: The final unit of this course challenges you to apply all the concepts learned to create a functioning AI agent that scores well on the GAIA benchmark, which you can learn about here.
This course will teach you both the theory behind AI agent applications and how to build them, culminating with a project that allows you to apply the skills learned.
To get a certificate for taking this course, there are two ways:
- Complete Unit 1 of this course for a “fundamentals” certificate
- Complete Unit 1 of this course, 1 assignment, and the final challenge for a “certification of completion.”
To get these certificates, you must complete the assignments before July 1, 2025. If you stumble upon this course after the deadline, you will still be able to access all the course material, although you will not get a certificate.
Summary
In this article, I have curated three generative AI courses at various skill levels that are a must in 2025:
If you only have time to take one course, I recommend the generative AI course by Microsoft.
Once you complete the generative AI course and want to learn specific topics like building AI agents or creating reasoning models like DeepSeek R1, I suggest taking the other two courses mentioned in the article.
 
 
Natassha Selvaraj is a self-taught data scientist with a passion for writing. Natassha writes on everything data science-related, a true master of all data topics. You can connect with her on LinkedIn or check out her YouTube channel.