Start Your Gen AI Journey: 3 Easy Beginner Steps!

Key Takeaways

Hello everyone, how are you? So many students have asked me that I want to start with my geni journey. I want to start with my LLM journey but they don't even know where to start. And this is a very common question. So what I have done is in the next few minutes I'm going to tell you how to start with your LLM journey. There are going to be three steps and you will have to look at my screen recording as well. So let's start. Step number one, you'll need to know about something called as Google Collab. C O L A B. You can Google it as well. And what will happen is so if let's just Google it together. Google Collab. Okay. What will happen is that it will lead you to and you can click on new notebook and it will lead you to the welcome page of Google cola. Now this is an absolutely amazing tool. You can do everything that you want in an expensive computer. You don't really need to buy an expensive computer. You have Google Collab and if you look at it, it will give you a lot of RAM, a lot of storage. I think it gives you 32 gigs of CPU RAM and 15 gigs of GPU RAM. This is not something that is easily available, but this is available for you for free. All you need to have is internet access and a Gmail account. If you have those two things and obviously a laptop, if you have those three things, you have a Google Collab account. Now the laptop can be fairly simple because all of the weightlifting all of the heavy weight lifting is done by Google collab itself. So let's just look at what you can do. So first you can create a notebook kind of a content. So all of the markdown that you can create in Jupyter notebook it is doable in Google Collab as well from including your heading two, heading one, heading three, your bold text, your italics text, your code links and even images. Okay, you can also have ordered or unordered list. Again, simple markdown. Yeah, but does not Google Collab does not stop there. It also gives you a free generation capabilities. It gives you a free Google uh GitHub copilot kind of a tool. So let's just click on uh generate and let's understand what we can generate with it. Let's for example create a simple pie chart using some sample data and it will generate the content for you. It will take some time maybe a few seconds but it will generate that content for you. So let's just wait and see what it spits out. Yeah. And all you need to do is accept and run. And let's just look at it. Yeah, it seems to be working perfectly fine. It's in fact absolutely amazing. You can have a a basically a repository of whatever you want of the chats and yeah, we were able to generate a simple pie chart. I've also tried to create other uh you know simple prompt based code for including reversal of lists. So it it generates really complex code as well and it including a diamond pattern as well. Again I did not do anything. I just wrote in English. I need to create a for loop with that that's print that prints out a diamond pattern. That's that's absolutely amazing. Again remember all of this is for free. Okay. Uh you can generate random prime numbers as a prompt and yeah. Okay. So Google Collab not only gives you a Python uh environment with capability to generate code but it also gives you a virtual machine. We all understand that in the real world Linux or shell is something is a skill that we all have to learn and Google collab provides you that it is not just a notebook it is a virtual machine as well. Now uh by default it gives you uh an Ubuntu release an Ubuntu flavor of Linux and again there are many many flavors of Linux but Ubuntu is the most popular one and that is what you get to create a shell based command all you need to do is add exclamation point before any command that you want right and then it will be a shell command okay and as I Google Google Collab also gives you a free GPU as well. As you can see over here, 15,000 MGB or 15 gigs of GPU RAM. This is available in really really expensive laptops. I have an expensive laptop and I have 16 gigs of GPU RAM only. I should not have purchased this laptop. Instead relied on Google Pollab. Okay. Okay. Not only that we I have seen students suffering from you know merge dependency hell. So dependency hell basically. So if they install numpy version let's say 1.8 uh pandas 2.0 will not work. If they in uh install mattplot lib seawn will not work. if they install OpenCV skarn will not work. So instead of you know playing around with in your local wasting time on uh identifying the best versions for your particular local environment, Google Collab provides you a stable latest environment. You don't have to install any of these including really large libraries for geni and deep learning as well. Obviously there's machine learning, there's visualization, there's uh deep learning and uh things like torch vision, lang chain. All of these you don't really need to install. These are available out of the box in Google Collab. You can even access external data. This may be in your Google Drive, in your Kaggle account, in your hugging face account and you can access that in in you know the the the regular way of Python coding. So that's not an issue as well. So that was step number one, set up something called as Google Collab. Step number two, you need to set your free LLM API key using something called as Google AI studio. Right? So all you need to do is click on a studio.google.com or you know you can Google a studio.google.com google.com and then uh let me just go to a yeah you can click go to aistudio.google.com and then you have to do uh go to the get API key at the bottom left. Once you click on that all you need to do is create an API key and it will ask you for a project. By default uh it is going to be the Gemini API project and you can just create an API key. So this API key should not be public. So what I'll do is I'll be deleting this particular public key. But yeah, all you need to do is copy this API key. Go to Google Collab again. There's a there's a key icon on the left. Click on this and you can set it in your uh in your collab secrets. So this is going to be your Google API key or Gemini API key. setting up setting it up in Google Collab. Okay, so this is pretty much it and let me just go and delete this. So whm as you can see it's completely free. Yeah, then all you need to do as as uh another step is install Google- geni library and that's it. Now you can use any LLM and that's it. Now you can use Google's Gemini uh really really and that's it. You can use Google Gemini which is a really really powerful LLM comparable to GPT5 GPT 4. All of those are comparable to G Gemini 2.5 Flash, Gemini 2.5 Pro, right? And let's just look at how to use it. Well, first you'll be importing some a few things. Obviously, you'll have to import from Google J from Google import jennai. Then you will import from Google Collab import user data. This is where you uh basically set up your uh Google's API key. So my variable for Google API key is Google API key. Let's just check. Yeah, I can just copy paste it. Right. And all you need to do is import these. Okay. Then geni.client API key equals user data that we imported, right? Get Google AP Google API key. That's that's all you need to do. Then you can start running your experiment. So for example, uh now this can these can be prompt engineering experiments. These can be rag based experiments. These can be agentic AI experiments. But yeah, all you needed to do was this. Okay. So let's look at a prompt engineering experiment. So how can Geeks for Geeks help a data science aspirant? What are some courses courses for data science they offer? So this is the prompt geni client that we defined above here. uh it have it provides a models subp package and within that sub package there's a function called dotgenerate content once you generate once you uh set it up all you need to do is pass the the model that you want we are going to use Google 2.5 uh Gemini flash and the prompt as contents and that's it let's just ask it to run it yeah so it's it's given a very detailed output in fact. But yeah, you can use any prompt. But then the question becomes how do you learn prompt engineering? Well, I have written an article called unified prompt framework or a unified framework for an effective prompt and you all you need to do is scroll to the bottom of this particular article. Okay. Copy paste this entire green gray area and let's go to chatgpd.com and I'm going to stay logged out and all you need to do is change your requirements. So you just have to describe what are the requirements that you have. So for example, I want to learn data science. Provide a road map. So it will give you an optimized prompt. Using this prompt, you can generate another optimized prompt which defines the role, the context, the task, the constraints, the examples, the output format, everything is provided. Okay. Yeah, I don't want to sign up. So all you need to do now is literally open up a new chart, copy paste the output of the optimization prompt optimization exercise and yeah that's it you just remove the extra fluff and maybe just say teach me uh about genac So it will generate an entire road map for you and you can do this programmatically with this strategy as well. So for example, let's just do instead of this particular prompt, let's just copy paste this prompt and Let's just run this. Let's see what will happen. It will take some time. Even I think chat GPT is still giving out some output, right? So this will also take some time. Yeah. So there is an output. So data science road map beginner to job ready. It gives you a lot of resources, a lot of links, a lot of uh topics. It divides everything into phases. It's really really amazing. Yeah, that's that's absolutely amazing. So this is how you can use an a free LLM. You don't have to pay money to set up your LLM. You don't have to pay money to set up your Google Collab. You don't have to pay money to learn how to do prompt engineering. Right? So, thank you for today. If you have any questions, please let me know. Also, I'll be starting a generative AI training program which has an absolute detail around Genai in abs, you know, get your career ready for Genai projects. So yeah, please do sign uh sign up to that and if you have qu any questions, please let me