Hi !
It’s Abdullah Fazal, 12 grade student form Pakistan , recently I started deep diving to learn agentic ai but not by using ready-made tools , becuase i am following a book which focuses on agentic ai but it has extensive chapters and all concepts which eventually make you a full stack AI engineer .. so i just completed core python concepts essential to become ai engineer for instance asyncio and core GIL related concepts for parallelism and OOP’s , data/metaclasses etc ….. I am also confused to some extent whether where this whole roadmap would lead me … for your information now i am gonna learn google adk for agents instead of clause and open ai agents sdk since they dont give you free tiers .
I want a professional person to give his opinion not on just my progress but also on my direction after analyzing the content of this book (since i cant type links so please type this on your browser “panaversity ai native book“)
Thankyou.
If it isn’t clear from the initial response, that isn’t what you received. Instead, you got text generated by a large language model. Neither a professional nor a person. Not someone who has direct experience with successful and unsuccessful projects, or hiring and firing people for projects. Not someone who has had a long enough career to know what is valuable over time, versus what is ephemeral and changes constantly. Is what you got reliable and useful? Perhaps, but just like any information provided on these fora, you have very little information on which to evaluate the credibility of the source. In the case of the LLM, it is generating words for you based on the numbers it has encoded in its tokenization and embedding modules. If I were you, I would try to find a few additional sources, especially that might understand your local employment market if that is a driver for your choices, and compare their responses. Ultimately there is no single path right for everyone, as it depends somewhat on what you are interested in and what you are good at. From reading quickly through your list, it seems there is some risk of learning a little about a lot of things, whereas hiring managers often require more depth in fewer things. Also depends on what you want to do with all this learning. Academic research? Work for a big company? Start your own? I would also encourage you to not forget about why all this shiny tech is important ; you didn’t mention anything about what you want to do with these capabilities, what business or social value you can create with them. Finally, don’t focus exclusively on specific current technologies and neglect understanding the role they fill in an overall solution architecture. Architecture, the major functional components and the interfaces between them, changes much more slowly than technologies implementing those components. Good luck on your journey.
2 Likes
Means a lot actually that you spend your valuable time on me… anyway the response from LLM was something i get everyday but the reality that you exposed in your response actually something that makes confusion crystal clear…. the main confusion still lies on me which is what i want to do or in which domain i have interest since many things are overlying to each other, sometime i think should i explore data side or stick to learn how to build agentic systems then i think if i can build agentic systems through tools eg n8n that people are actually building and making money, then why to learn these technologies eventually i came to take decision that i must follow this book or roadmap since its going to give me essential skills which i need to work in ai domain as an engineer although learning is specifically focused on agentic systems but my skills will be applicable as a general ai engineer or in future i can specialize in one specific domain like in deployment section as this book is covering all domains which one is not supposed to master right? what your taughts do i have to first touch all concepts then deep dive into one specific domain mastery?
I know this text is getting long enough but i want to tell you my next 2 months plan … i will start building a production ready digital fte or you can say multi agent system which will provide 2 purpose .. one to add a solid project in my portfolio … secondly to have a production ready system which actually solves a real problem and can be monetize ..
In order to do so, as per now i have learned MCP and core adk concepts which i will use to implement to build agents or maybe i will build through scratch.. “no need for adk or sdk if you think production ready system shouldnot depend on adk?”… first i will complete my core concepts of part 5 of book which is about rag , databases, TDD, agentskills , Building MCP Servers, Fast Api and some more concepts then i will parallely start filtering a niche and eventually plan to build a product which i dont know so far what it would be more probably a customer support Digital FTE … so i was telling you my approach that during learning parallely i will start building this by breaking this into milestone and implement new things incremently as i learn simultaneously. and after lets say 1 month of dev when my product will be ready at that time it is possible that i will have knowledge of deployment section enough so that i can deploy my product and distribute it if possible otherwise it will just serve as a solid project inmy portfolio.
I want your last response on it in which i want you to give me details guidance and your thaughts so far and loop holes on my plan becuase i think you got my main confusion which is in my heart.
Building any production-ready system in 2 months strikes me as quite ambitious. However, I understand that, sometimes, grand ambition leads to grand achievement. If that is your objective, who am I to discourage you?
I will suggest that you look for every opportunity to leverage existing components and not write common capabilities from scratch when it isn’t going to be a differentiator for your product. Focus instead on a new combination or new end user value. If building a product yourself is your goal, I would start there and decide what it will be, then work backwards to the capabilities needed. For each capability you can then make a better decision about the set of skills and knowledge you need to acquire and whether to build from scratch versus leverage toolkit or library. Hope this helps