{
  "video_id": "ki0UR5ZP_QY",
  "channel_slug": "techwithtim",
  "channel_handle": "techwithtim",
  "title": "Do THIS Instead of watching endless AI Engineer Roadmaps (DataCamp Review)",
  "duration_seconds": 935.0,
  "url": "https://www.youtube.com/watch?v=ki0UR5ZP_QY",
  "upload_date": "",
  "transcript": "If you want to become an AI engineer,\nthen watch this video because I'm going\nto save you months of time. And that's\nbecause there are hundreds of different\nAI engineering road maps out there. It\nfeels like they're changing every single\nday and they mention literally thousands\nof skills and if you follow the wrong\none, you could be spending months\nlearning something that doesn't actually\nhelp you land a job. So, in this video,\nI'm going to share with you a road map\nthat I personally went through. One that\nI spent 20 hours learning. So, I\nactually went through the topics, I did\nthe projects, I did all of the\nexercises, and I can honestly recommend\nit because I've gone through it myself\nto test and verify the criteria. Now,\nit's not perfect, but it's one of the\nbest ones that I found, and I'm going to\nshare with you an honest review here and\ngo through it step by step. So, even if\nyou don't want to purchase this, because\nit is a paid course that I'm going to\ntalk about, you can at least follow the\nskills in the curriculum that are inside\nof this road map. And again, I know that\nit's a good road map cuz I actually went\nthrough it, spent over 20 hours doing\nit. And I feel like I've genuinely\nlearned a lot even though I'm already a\npretty experienced AI engineer. Okay, so\nthat said, let's get into the video. Let\nme share with you the resource that I'm\ntalking about. I also want to discuss\nwhy you might want to become an AI\nengineer because it's a really hot topic\nin 2026. So, first, let me just give you\na quick primer on what an AI engineer\nactually does because that's really\ngoing to dictate the road map that\nyou'll want to follow. Now AI\nengineering is a lot less about training\nmachine learning models or doing you\nknow statistics and mathematics and much\nmore about actually applying things that\nhave already been built. So now\nespecially in 2026 it means using\npre-trained models it means using LLMs.\nIt means creating pipelines using rag\nretrieve augmented generation and\ndeploying these applications into\nbusinesses or you know into companies\nwhere you're actually working. So when\nyou're looking at road maps here, you\nwant to look at things that are very\npractical, that are giving you a lot of\nprojects, and that are teaching you\nagain the implementation rather than all\nof the theory. You honestly don't really\nneed to know any math at all to be an AI\nengineer. And again, what you're doing\nis you're using things that have already\nbeen created, like a hugging face model\nor an existing LLM to actually create\napplications that deliver real value.\nNow, because of the practical nature of\nthis role, this has quickly become one\nof the fastest growing jobs for software\nengineers. And it's one where we're\nactually seeing an increase in demand.\nWe're seeing massive salaries. You have\nstarting pay, you know, over $150,000\nper year, whereas some of the other more\ntraditional dev jobs are a little bit\nmore stagnant or even declined. So,\nanyways, that's a little bit about AI\nengineering. Now, I want to dive into\nthe road map. Now, full disclosure here,\nthe road map that I'm going to share\nwith you is a paid resource. You can\naccess parts of it for free, so you can\ntry it out. And of course, you can view\nthe entire curriculum and it comes from\na long-term partner of this channel,\nData Camp. Now, I know immediately a lot\nof you are going to throw the red flags\nup in the air, and I completely\nunderstand. But what I want to mention,\njust to get it out of the way, is that\nI've been working with Data Camp now for\n3 plus years. I genuinely really enjoy a\nlot of their courses. I've gone through\nmany of them myself, and I started using\nthem well before they ever were a\nsponsor of this channel. So, while yes,\nthey do sponsor this channel, I do get\npaid from Data Camp, this is an honest\nreview and recommendation of this\ncourse. Again, I went through it myself.\nYou're going to see I've completed all\nof the different things. And while it's\ngreat and I'm recommending it, it's not\nperfect. And I'm also going to share\nwith you the negatives of this course in\nthis video. So, anyways, you guys can\njudge it however you want. But this is\none that again I genuinely do recommend.\nI went through myself and there's a\nreason I've partnered with Data Camp for\na very long time on this channel. Okay.\nSo, first let's just get the pricing\nright out of the way. Now, you don't\nhave to pay for one individual course\nhere. If you want to access this, what\nyou'll need to do is pay for either a\nmonthly or yearly subscription with Data\nCamp. You can see that if you have the\nbasic subscription, which is free, you\ncan access parts of the course. There's\na few free modules that you can go\nthrough. And then if you upgrade to\npremium, which is the one that you would\nhave as an individual, you can pay $13\nper month and get access to all of the\ndata camp courses, not just the one that\nI'm going to show you in this video.\nOkay, so that's the pricing. And I do\nhave a discount link. I'll leave it in\nthe description if you want to get 25%\noff any of the different courses. Again,\nyou can access all of them. It's just a\nsubscription that you pay for it and you\ncan cancel whenever you want. Okay, so\nthe course that I personally went\nthrough here that's going to contain the\nroad map we're going to have a look at\nis the associate AI engineer for\ndevelopers. Now, this is not designed\nfor complete beginners and it assumes\nthat you have some Python fundamentals,\nyou know a little bit about programming\nand that you've built some things in the\npast. If you're an absolute complete\nbeginner, there's some better tracks\nthat you can go through. This is really\nmeant for those of you that have the\nbasics down, you know your core Python,\nand now you're looking to actually learn\nthose kind of AI engineering skills. So,\nI'm going to go through the curriculum\nhere so you can see exactly what's\ninside of this course. It's actually\nnine courses kind of combined in a track\nor road map, hence the title of this\nvideo. It has three projects and it\nestimates that it would take you 26\nhours to complete. For me, I was able to\ndo it a little bit shorter cuz I'm\nalready familiar with some of the\nconcepts. But if you're a complete\nbeginner in AI engineering, this is\nprobably a pretty good estimate. Now,\nthis also comes with a data camp\ncertification for this course. I haven't\ndone this one, but there is one that you\ncan do that we'll look at later. And\nanyways, what I want to do is start\ngoing through the different topics. So\nthe first thing that's included here is\ncalled working with the OpenAI API. Now\nwhat this does is show you how to\nactually send requests to OpenAI so that\nyou can get responses back from the\nmodel. It talks about things like model\nselections, using the response API,\ndifferent types of messages like user\nmessages, assistant messages, system\nmessages, and generally just shows you\nhow you can actually interact with\nthirdparty LLMs or again like OpenAI,\nyou know, API and get responses back.\nNow, I'll go through some individual\nmodules in a second, but I believe this\none is free that you can try out if\nyou're just looking at this. Okay. Next,\nwe have prompt engineering with OpenAI.\nNow, from here, what this does is go\nthrough all of the different prompting\nstrategies. Talks about oneshot\nprompting, multi-shot prompting, giving\nexamples, different ways that you can\ncraft prompts to get better results,\nstructured outputs. And honestly, this\none for me actually taught me quite a\nbit about prompting that I didn't know\nbecause it's very thorough, and there's\na ton of different exercises where you\nactually have to write your own prompts\nand pass the test cases that data camp\nhas. Now, after that, you immediately\nget into a project. This project is\nsimply just triggering the OpenAI API,\nhaving multiple conversations, storing\nthe previous conversations so that the\nmodel kind of has context and then\nkeeping track of like the number of\ntokens, the cost, all of that kind of\nstuff. Now, after that, it moves to\nworking with hugging face. Now, again\nguys, if you don't want to take this\ncourse, just follow these concepts on\nyour own. You can learn this stuff for\nfree. It's just not going to be in one\nplace with the exercises. Okay, so\nHugging Face, what is Hugging Face?\nWell, Hugging Face is actually a\nplatform that has a ton of open-source\nmodels that you can use for free. So,\nwhat this starts to do is teach you how\nyou can utilize those different models\nto do things like text classification,\nimage, image classification, all\ndifferent kinds of machine learning\ntasks where you're not building the\nmachine learning model yourself, but\nyou're taking it from hugging face,\nyou're combining it into some type of\npipeline, you're passing in the correct\ntype of data, and then you're getting a\nresponse back. There's actually quite a\nfew things in here that are pretty\ninteresting. So, for example, if I go\ninto here, I'll just show you a few\nexamples of kind of how it's structured.\nSo, if we look here, you can see we have\ngrammatical correctness, tokenization,\nusing auto classes, extracting text with\nPIP PDF, building a Q&A pipeline. And if\nI just go into maybe one of the examples\nhere, okay, let's go next to the next\nvideo. You can see actually this one is\num just classification. Let's go into\none that's coding. You can see that\nyou're actually pulling in, you know,\npre-trained models. You're using a\ntokenizer to tokenize text. And you have\nto go through these exercises yourself\nand pass all of the different\nsubmissions. And of course, they get\nmore complicated as you go through the\ncourse. So that's kind of the basics on\nhugging face. Let's go back to the\ncourse content. Okay. Now, after hugging\nface, it moves to LLM ops. Now, this\nsection is not super detailed. It's more\nkind of a highlevel explanation as\nopposed to actually getting into like\nDocker and Kubernetes and the DevOps\nstuff that you would do in LLM ops, but\nit gives you some information on what\nLLM ops actually is, how it differs from\nMLOps because these are two kind of\nseparate categories and explains things\nlike the development phase, the\noperational phase, and different things\nyou want to consider when you're picking\nmodels or picking a database or should\nyou go with an open source model or\nclosed source model. should you\nfine-tune a model and gives you kind of\nsome working understanding as you start\nmoving into some more complex topics in\nthe course. Next, developing AI systems\nwith the open AI API. So, this is where\nwe're kind of combining things like\nhugging phase as well as open AI and\nbuilding some larger applications. This\nis where it starts to get a little bit\nmore complicated and picks up in terms\nof kind of the speed at which you're\nlearning. Then we move to embeddings\nwhere we start getting into vector\ndatabases, talking about rag and\nbuilding practical applications with AI.\nIf you're not familiar with rag, this is\nretrieval augmented generation. And this\nis a really popular technique. It's used\nlike almost all the time now by AI\nengineers to be able to take large\npieces of data like a database or a text\ndocument, feed it into an LLM, get\nrelevant information from it, and then\ngive you a response that's based in\nfacts. So, it starts going really in\ndepth into what embedding models are,\nhow you embed text, how you create\nvector databases. Again, this another\nproject. And then we start getting into\nother types of databases because this\none I believe uses like Chroma DB. And\nthen we start looking at Pine Cone which\nis a little bit more advanced. It's just\nanother type of database. I'm not going\nto go into it too much. Then we get into\nsoftware engineering principles in\nPython and developing LM applications\nwith lang chain where we start again\ncombining these creating some more\nindepth chains that are a little bit\nmore complicated have AI agents have\ntool calling and really kind of building\nsomething that's a bit more advanced\nthan simply triggering an API or simply\nusing a hugging face kind of transformer\nor pre-trained model. Now, I'm not going\nto go through every single section\nextremely in depth, but generally\nspeaking, I felt that this was a really\nstrong kind of introduction to AI\nengineering, especially at the price\npoint of like $13 to $20 per month,\ndepending on how you pay this. And while\nI don't feel like you would be a, you\nknow, complete AI engineer when\nfinishing this course, it at least gives\nyou a lot of those core topics and a lot\nof that understanding to be able to very\neasily branch off, work on your own\nprojects, and learn things that are a\nlittle bit more complex. And I believe\ndata camp actually has some tracks that\nare beyond this that you can take\nafterwards that go into some more\nadvanced because for example things like\nLLM concepts it is just concepts right\nit's not actually teaching you like all\nof the Docker Kubernetes and some other\nthings that you'd probably want to learn\nbut generally speaking I think you would\nbe very well set up to then go on to the\nmore advanced topics which would help\nyou get employed. Okay, so that's the\nhighle road map. It progresses very\nwell. At the beginning I will say that\nis a little bit slow. If you're someone\nwho's a bit more advanced like me,\nyou're probably going to find yourself\nskipping through a few of the lessons\nbecause it may teach you some things\nthat you've already done before, like\nI've already worked with the OpenAI API,\nfor example. But I want to show you how\nthe course kind of progresses because\nthis is what I like about Data Camp\ncompared to a lot of other course\nplatforms out there. So, if I click into\none of the lessons here, you can see the\nfull outline up here, which is always\nnice. They have this XP thing, which I\ndon't really find that valuable, but\nwhatever. They have like kind of some AI\nchat. They have the transcript and then\nall of the videos that you see here are\nbetween like three to five minutes long.\nSo what I like is that you're learning a\npretty quick concept. You're kind of\ngetting information. It has like lots of\ncharts and graphs and you know whatever\nthe PowerPoint slides that they have\ngoing on here. So it's very visual and\neasy to understand. And then as soon as\nyou kind of have grasped the concept\nfrom this uh you know presentation or\nwhatever they have you immediately go\ninto some kind of exercise where you\nactually need to answer a question or\nyou have to you know write some code. So\nyou have four or five minutes of\nlearning where you're kind of watching\nthis. I was watching a lot of the videos\nin like 1.5 or 2x speed. The instructors\nare different for all of the courses but\nit's you know very consistent in terms\nof the structuring and then you just\nimmediately get tested. So you're\nspending probably 80% of your time in an\nenvironment where you're writing code,\nyou're doing exercises, you have to look\nback at the video, you have to find\nwhatever the result was. So you're\nactually being challenged and you kind\nof have this interactive learning\nplatform which is a lot more effective\nfor keeping the concepts and retaining\nthe information compared to just\nwatching like 20 or 30 minute long\nvideos. So it's really focused around\nthis kind of like interactive\ncloud-based environment which you can\nsee here where you can run the code, you\ncan submit the answer, right? and you\ncan get hints and see what it is that\nyou're supposed to do. So, I'm just\nsharing that with you because that's the\nlearning style that I personally like\nand that I find is the most effective.\nAnd all of the different kind of\nsubcourses or modules all follow that.\nSo, you'll see like wherever it shows\nthis little code block. That's a coding\nexercise. And then you'll see a few\nvideos here in this one there's like 1 2\n3 four videos. The rest are exercises.\nSo, you're spending most of your time in\nthat code editor environment which is\nwhere I find the learning actually\nhappens. Okay. Okay, so that's the\noverall kind of structure and how the\ncourse works. Now, I will mention that\nthere is also a certification that you\ncan do afterwards. They have a bunch\nactually on data camp and this is going\nto be like a timed proctored exam where\nyou actually have to go through answer\nall these questions. I haven't actually\ndone this uh certification yet, so I\ncan't speak to how difficult it is, but\nI like that they have those and then of\ncourse you can add it on your LinkedIn,\nput it on your resume, etc. Just want to\nmention they do have that available.\nOkay, now I want to give you my kind of\nhonest review of this, right? So, first\nthe road map itself I do feel like is\nreally solid. Like I mentioned, it's\nreally good for people that already know\ncore Python and that are a little bit\npast beginners and just want to start\ndabbling into AI engineering. Now, I\ndon't feel that this covers everything\nthat you would need to know to become an\nAI engineer. However, I think it's a\nreally good start. I think if you get\nthrough this, then it's going to give\nyou some really solid context, which\nlike I mentioned is going to make it way\neasier to go into the more advanced\ntopics that you would need to actually\nland a job. And this gets you probably\nhalfway there, which I think is a good\nstarting point. And again, I felt even\nas an experienced engineer, I learned a\nlot going through this. Now, what I will\nsay is that there are a few sections\nhere, for example, like the LM ops\nconcepts that I wish were a little bit\nmore detailed. I know it's meant to be\nkind of high level and not overwhelm you\nwith information, but I wish that we\nactually looked at some examples of\ndeployments. We did, you know, a quick\nDocker container. We actually deployed\nsomething out because I don't find it's\nthat useful just to have the theory. I\nthink it's better if we actually dive in\nand do some of the exercises because\nthis one was not super detailed. I will\nalso say like I mentioned before, it is\na little bit slow at the beginning. So,\nthere are a lot of kind of repetitive\nthings that it's getting you to do and\nthen it increases in difficulty pretty\nsharply fairly fast. So, that is one\nthing to keep in mind that it's going to\nfeel probably pretty easy at the\nbeginning and then as soon as you kind\nof get over here to where we started\ntalking about embeddings and vector\ndatabases, it does increase quite a bit\nin difficulty. not meaning you can't\nfollow along, but it is just kind of a\nstark contrast from the start of the\ncourse. Other than that, the projects\nare quite good. I wish they had another\nproject at the end just to wrap\neverything up and combined a lot of the\ntopics and maybe give us something a\nlittle bit more complex. Then that's\noverall kind of my review and feedback\nfor this road map. So anyways guys,\nthat's going to wrap up this video. I\ntotally understand if you watch this and\nyou think I'm completely biased because\nI work with data camp, but I'm genuinely\ntrying to give you my honest review of\nthis. I spent a lot of time going\nthrough the course content. You can see\nI've completed it all here just so that\nI can give you an honest, genuine review\nof the platform. It's not perfect. They\ndidn't do everything 100%. Again,\nthere's a few negatives here and like I\nsaid, this isn't going to completely\nmake you an AI engineer. But if you are\nlooking for a really affordable, great\nplace to start that's designed again for\npeople that already know that Core\nPython, this is one of the best\nresources that I found. I have\npersonally tested and tried it and I can\nsay that it is very high quality and\ngoes through a lot of the content in a\nway that's very digestible and again\ninteractive to learn which is why I'm\nrecommending it and I don't think you\nwould regret checking it out. So anyways\nguys if you enjoyed this video leave a\nlike, subscribe and I will see you in\nthe next one.",
  "transcript_chars": 19927,
  "ingested_at": "2026-06-17T04:32:56.454356+00:00",
  "source": "channel",
  "yt_meta": {
    "view_count": null,
    "like_count": null,
    "channel_id": null,
    "categories": null,
    "tags": null
  }
}