Best Roadmap to Learn Generative AI in 2024

Yana Khare 25 Apr, 2024 • 6 min read


In the ever-evolving realm of Artificial Intelligence, Generative AI stands as a beacon of innovation, continually pushing the boundaries of creativity and intelligence. As we stride into the promising landscape of 2024, the pursuit of harnessing Generative AI’s potential beckons enthusiasts, researchers, and practitioners alike. This article delves into the intricacies of the best roadmap for Generative AI in 2024, charting a course through the dynamic advancements, emerging trends, and transformative applications that define this cutting-edge field.

Join us on a journey that unravels the key milestones, tools, methodologies, and insights, offering a comprehensive guide to navigate and excel in the realm of Generative AI in the year ahead. Whether you are a complete beginner in AI or a working professional such as a Data Scientist, Machine Learning Engineer, Deep Learning Engineer, or any similar role, this learning path will equip you with the skills and knowledge to master Generative AI.

So, fasten your seatbelts and prepare for an exhilarating journey into the world of Generative AI!

Roadmap: How Can I Start Learning Generative AI?

You can start learning Roadmap for Generative AI through 4 personas: User, Super User, Developer, and Researcher. We will discuss each persona in detail. Before moving ahead, you need to be familiar with terms like Generative AI and Foundation Models.

  • Good understanding of the Generative AI and Foundation models and their infinite use cases.

Let’s explore different personas now.

generative ai learning path,roadmap for generative ai

1. User

There is no better way to learn Generative AI than experiencing it. The first persona is to become a user of the Generative AI tools. Sign up and create an account on any of the Generative AI tools and gain hands-on experience. Get familiar with these Generative AI tools, understand what they are, know their capabilities and features, and experiment with them.

Now, we know better the pros and cons of Generative AI tools and how they can help us in our work. The next step is to go in deeper and understand how to use it effectively.

2. Super User

After gaining hands-on experience with the Generative AI  tools, the second step is to improve our knowledge and learn to use the tools better.

Generative AI tools have a lot of potential, which has not yet been explored. We need to learn to apply the right techniques to use them effectively. Most Generative AI tools generate responses based on the natural description known as prompt. Prompt writing is an art. We need to learn about prompt engineering in detail to explore Generative AI to its full potential. Here’s what you need to do for it:

  • Learn about Prompt Engineering.
  • Explore the best and most effective prompts for using the Generative AI tools.
  • Understand the best practices for writing prompts.

3. Developer

Now, we are comfortable using Generative AI tools effectively. The next phase is learning how these generative AI models work and finetuning these models on our datasets.

You need hands-on experience with machine learning and deep learning to do that. I recommend reviewing the prerequisites below before starting with machine learning and deep learning. Feel free to skip the prerequisites if you are already comfortable.


  • Good understanding of Probability and Statistics concepts.
    • Probability: Probability, Conditional Probability, Bayes Theorem, etc.
    • Statistics: Normal Distribution, Central Limit Theorem, etc.
  • Good understanding of Linear Algebra concepts like vectors, matrices, and systems of linear equations.
  • Good knowledge of Calculus concepts like gradients, derivatives, and partial derivatives.
  • Hands-on experience with programming languages like Python/R.

3.1 Machine learning

  1. Comfortable with supervised and unsupervised learning algorithms like linear regression, logistic regression, random forests, k means, etc.
  2. Know to build machine learning models on tabular datasets.

3.2 Deep Learning

  1. Good understanding of deep learning architectures like Multi-Layer Perceptron, Recurrent Neural Networks (RNNs), Long Short Term Memory models (LSTMs), Gated Recurrent Units (GRUs), and Convolutional Neural Networks (CNNs).
  2. Have hands-on experience with at least one of the deep learning frameworks like Keras, Tensorflow, Pytorch, or FastAI.
  3. Be able to train deep learning models using any of the deep learning frameworks mentioned above. For example:
    • Train Multi-Layer Perceptron on the tabular datasets.
    • Build RNNs and CNNs for unstructured data, i.e., text and image.
  4. Knowledge of pretrained models for image data and their different types. For instance, know how to finetune them on the downstream tasks.
  5. Learn about language models and build them with LSTMs/GRUs.
  6. Gain knowledge of Attention Mechanisms and know the limitations of using LSTM for working with longer sequences.
  7. Understand the architectures of Autoencoders and GANs and be able to train these models on our datasets.

3.3 Generative Models for NLP and Computer Vision

Now, you can customize your learning path depending on your interest. If you want to learn and build Generative models like ChatGPT, you can choose the generative models for NLP. If you are interested in building models like Midjourney and DALL-E 2, you can select generative models for computer vision.

You Can Also Watch Roadmap for Generative AI:

3.3.1 Generative Models for NLP

If you choose NLP as your area of focus, the following learning path will guide you to mastery of Generative Models for NLP.

  1. Discover the power of Large Language Models (LLMs), the foundation models of Natural Language Processing (NLP).
  2. Learn about popular LLMs like Transformers, BERT, GPT 3.5, PaLM 2, and many more.
  3. Understand how to use Large Language Models (LLMs) for downstream tasks: Finetuning and In-context learning i.e., Zero-shot, one-shot, and few-shot learning.
  4. Uncover the best practices for training LLMs, including challenges, scaling laws, and efficient training mechanisms like parallel and distributed architectures.
  5. Learn how to pretrain LLM on your domain-specific data.
  6. Understand and implement different techniques to fine-tune LLM for downstream tasks.
  7. Learn different optimization techniques to accelerate model finetuning like Adapters, LoRA, QLoRA, etc.
  8. Know LLMops: How do you deploy LLM in production?
  9. Explore cutting-edge models like ChatGPT and BARD and understand their training process, including concepts like Reinforcement Learning from Human Feedback (RLHF), Supervised Fine Tuning, and Prompt Engineering.
  10. Know how to finetune ChatGPT on your dataset.
  11. Get hands-on with LLM frameworks like LangChain, AutoGPT, Vector DBs, etc.

3.3.2 Generative Models for Computer Vision

If you choose to delve into computer vision, this learning path will guide you in mastering generative models for computer vision.

  1. Learn about foundation models in computer vision: diffusion models and their different types.
  2. Understand how to finetune diffusion models for downstream use cases.
  3. Learn about stable diffusion models, including model architecture and training process.
  4. Learn how to finetune stable diffusion models on your datasets.
  5. Explore Mid Journey, DALLE 2, and any other similar models.

4. Researcher

The last stage is intended for researchers. To build your career in Generative AI research, you need to understand how to build these generative models from scratch. You should be well-versed in various concepts and techniques to build these generative models.

To be a researcher in NLP, you need to:

  1. Learn and implement attention models, Key Query Value (KQV) attention, layer normalization, positional encoding, etc.
  2. Get hands-on experience building your own GPT architecture from scratch.
  3. Understand the workings of reinforcement learning algorithms from basics to advanced.
  4. Learn about Proximal Policy Optimization (PPO).
  5. Implement RLHF from scratch.
  6. Build ChatGPT from scratch
  7. Stay updated with the current trends and research in Generative AI for NLP

To continue research in Computer Vision:

  1. Build diffusion models from scratch.
  2. Learn how to implement stable diffusion from scratch.
  3. Stay updated with the current trends and research in Generative AI for Computer Vision.


As we draw the roadmap for mastering Generative AI in 2024 to a close, this journey has illuminated the diverse pathways available to enthusiasts, researchers, and professionals eager to delve into the realms of creativity and intelligence. The personas of User, Super User, Developer, and Researcher serve as guiding lights through this transformative expedition, offering tailored routes for various levels of expertise and aspirations. This comprehensive roadmap for Generative AI charts a course that aligns with the evolving landscape of artificial intelligence, providing a structured guide for those navigating the exciting intersections of technology and creativity.

Remember, this roadmap for Generative AI is not just a linear path; it’s a guidepost offering flexibility, adaptability, and room for exploration. Embrace the challenges, engage in continuous learning, and stay attuned to the evolving trends in Generative AI. As 2024 unfolds, this roadmap stands as a compass, guiding you toward mastering the art and science of Generative AI, unveiling new vistas of creativity and innovation in the year ahead.

Yana Khare 25 Apr 2024

Frequently Asked Questions

Lorem ipsum dolor sit amet, consectetur adipiscing elit,

Responses From Readers


Prabir 29 May, 2023

Hello Arvind, thanks for the step by step guide. This article is now my one stop guide for becoming an LLM Developer/ Researcher. Can you please guide me to some useful resources for - LLMOps, best practices for writing effective prompts, fine tuning LLMs on my own domain specific data and building and training my own ChatGPT like model?