Building intelligent opensource Ai agents from scratch

In this talk I will explore the evolution of AI agent architectures using opensource language models and tools. Starting with basic API interactions, we'll progressively build toward sophisticated AI agents capable of reasoning, planning, and using external tools.

1. I will be teaching students how to make basic ai agents from scratch using python.

2. Difference between traditional automation and latest ai agents.

3. designing system prompt for such ai agents

4. using different llms to make these ai agents like gemini, llama, mistral etc.

5. i will be making use of the opensource library by openai to make such agents.

6. Basically ai agents are just llms with hands and legs to do tasks assigned to them and think before performing such tasks and can decide what to do in case of some block.

1. how to build ai agents from scratch

2. How to use openai library to build such agents

3. How to make use case specific ai agents

4. Using advance python to make them

5. Understanding the transformer architecture behind llms

6. Practical knowledge on these things with little theory to keep things interesting and keep the audience engaged.

7. How do the LLM's work behind the scenes like conditional probability concepts and some more life adjusting softmax and other things.

8. Different types of prompting techniques.

9. defining tools and letting the llm decide on what tool to use and when. I mean how one can implement these things.

10. After the talk, the students will be able to make ai agents that can easily automate their day to day tasks as well as keep them relevant in the job space which is being acquired by ai agents very fast.