AI Engineer Roadmap: Built From 40,000 Real Job Ads — Transcript

Full transcript

1. 0:00I've spent two months building an AI
2. 0:01system that runs daily and collect every
3. 0:04data and AI job ads. And now I use it in
4. 0:06order to understand how the job market
5. 0:08is moving. So now, instead of just
6. 0:10analyzing 10 jobs, the system collected
7. 0:12over 40k jobs from more than [music]
8. 0:1510,000 companies in 100 countries. And
9. 0:19more than 4,000 of them are actually
10. 0:21searching for AI engineers. Now, the
11. 0:23thing is about AI engineering, it is a
12. 0:24new role and the job market and the
13. 0:26platforms are changing so fast.
14. 0:28>> It all happens so fast. So fast.
15. 0:32>> So that it is really hard to find one
16. 0:34road map for AI engineering where
17. 0:36everyone agrees on. And most of road
18. 0:38maps are actually opinions. And my
19. 0:40favorite quote, "Without data, you are
20. 0:41just another person with an opinion." So
21. 0:44now, in this video I'm going to show you
22. 0:45what is exactly the job market, what are
23. 0:47the companies are expecting from you as
24. 0:49an AI engineer, everything, the skills,
25. 0:51the tools, the salaries, the real tasks.
26. 0:53And not only that, we can as well cover
27. 0:55the new trending AI role, the forward
28. 0:57deployed engineer. So that at the end
29. 0:59you will get a realistic road map that
30. 1:01is totally based from what companies are
31. 1:03looking for. And by the way, if you are
32. 1:05new here, welcome. My name is Baro and
33. 1:06I've spent 17 years in data engineering.
34. 1:09I lead one of the biggest data platform
35. 1:11project at Mercedes-Benz. And now, on
36. 1:13this channel I'm sharing everything that
37. 1:15I know and my experience. So subscribe
38. 1:17so that you don't miss anything. Now,
39. 1:18let's deep dive.
40. 1:20>> Your journey begins now.
41. 1:23>> All right. Now, let's start with this
42. 1:24question, what is an AI engineer? A
43. 1:26software engineer who builds
44. 1:27applications using AI models. And there
45. 1:30is a little bit of as well data
46. 1:32engineering because they go and build
47. 1:34pipelines to move and feed the data into
48. 1:36their AI system. Sounds simple, right?
49. 1:38But once you start searching for AI
50. 1:40jobs, you're going to find a lot of
51. 1:41chaos. Now, the biggest group of jobs
52. 1:43going to be for AI engineers and it
53. 1:44comes with different names, applied AI
54. 1:47engineer, gen AI engineer, generative AI
55. 1:50engineer, agentic AI engineer, different
56. 1:53names for the same job. And then comes
57. 1:55smaller groups like the AI researcher,
58. 1:58the AI architect, the AI product
59. 1:59manager. Of course, we don't forget
60. 2:01about the new trending one, the forward
61. 2:03deployed engineer. It is small, yes, but
62. 2:052 years ago, nobody used this title. Now
63. 2:08comes the most important question, what
64. 2:10are the skills that are expected from
65. 2:12you as an AI engineer? Now, in order to
66. 2:14answer this, I analyzed the section what
67. 2:16we are looking for. Of course, if you
68. 2:18look at the job description, you're
69. 2:19going to find like a long list of
70. 2:21skills, but don't worry about it. Most
71. 2:22of them are wish list. Nobody can
72. 2:24fulfill everything. And usually the most
73. 2:26important skill or the must-have going
74. 2:28to be at the beginning. What the data is
75. 2:30saying, the most important skill is, no
76. 2:32surprise, Python. The whole work going
77. 2:35to be depending on your knowledge of
78. 2:36using Python. Without it, you cannot do
79. 2:38anything. Now, the second skill going to
80. 2:40be the LLM, as well very frequently
81. 2:42mentioned in the job ads. So, those are
82. 2:44the top two, Python and LLM. And with
83. 2:46that, we are saying you are an engineer
84. 2:48that can use large language model. Now,
85. 2:50of course, we have long list of many
86. 2:53other skills, but I've grouped them into
87. 2:55groups. The first one is the programming
88. 2:56language, and as you can see, Python is
89. 2:58a clear winner here. And tiny number of
90. 3:00ads are asking for TypeScript and SQL.
91. 3:03Now, of course, obvious, my
92. 3:04recommendation is to master Python. And
93. 3:06if you want second one, then focus in
94. 3:08SQL. So, things are clear here. Now, to
95. 3:10the second group is about the AI skills.
96. 3:12Now, here we have a lot of names. Of
97. 3:14course, on the top going to be the LLMs,
98. 3:16then AI agent and drag. Those are the
99. 3:18core three that you need for the job.
100. 3:21Now, quickly, the LLM is the model that
101. 3:23you are going to use inside your
102. 3:24application. With an agent, you connect
103. 3:26that model to tools so that it take
104. 3:28actions. And drag, it is retrieval
105. 3:30augmented generation. Your application
106. 3:32search the company's documents and gives
107. 3:34back the relevant information to the
108. 3:36model before it answers. Now, by looking
109. 3:38to the next skills, this going to answer
110. 3:40question that a lot of people ask or
111. 3:42think of. In order to become an AI
112. 3:44engineer, I have to go and start
113. 3:45training models. I need a PhD, and I
114. 3:48have to understand all the algorithms
115. 3:50and the math. Well, no, my friends, you
116. 3:51are not a data scientist. You are not
117. 3:54doing machine learning and all those
118. 3:55stuff. Yes, they are sometimes mentioned
119. 3:57in the job, but this is not your main
120. 3:59job. You are an engineer. You have to
121. 4:01build an AI system and not a scientist
122. 4:03where you're going to go and start
123. 4:04training models from the scratch. And
124. 4:06now, the question is where I'm going to
125. 4:07learn all those skills to become an AI
126. 4:09engineer. I have partnered with DataCamp
127. 4:10because here's the thing, you remember
128. 4:12only tiny fraction of what you passively
129. 4:14watch. But when you learn hands-on,
130. 4:16writing real code, building real
131. 4:18applications, doing projects, retention
132. 4:21going to jump close to 80 and 90%. And
133. 4:23this is exactly the mindset of DataCamp.
134. 4:25And by the way, you write the code
135. 4:26directly in your browser and get instant
136. 4:28feedback. Now, if you are starting from
137. 4:30zero, begin with the Python programming
138. 4:32fundamentals track, four courses around
139. 4:3516 hours, no prior knowledge is
140. 4:37required. It covers the variables, the
141. 4:39data types like the list, the
142. 4:40dictionaries, the functions, the
143. 4:42packages, all hands-on with a project at
144. 4:45the end. Then after that, I recommend
145. 4:46you to take the Associate AI Engineer
146. 4:48for Developers track, 10 courses around
147. 4:5130 hours. It covers LLMs, AI agent, the
148. 4:54rag, all in one track. So, you are going
149. 4:56to build chatbots, semantic search
150. 4:59engines, recommendation system using the
151. 5:01OpenAI API, Hugging Face, LangChain, and
152. 5:04the vector databases. And as well, by
153. 5:06the way, it covers the LLM ops so that
154. 5:08you learn how to deploy and maintain
155. 5:10those applications. This is going to be
156. 5:11your daily job as an AI engineer. And at
157. 5:14the end, it prepares you for the
158. 5:15DataCamp AI Engineer, so you don't just
159. 5:18learn the skills, you can prove them.
160. 5:20So, my friends, both tracks in one
161. 5:21subscription. And by the way, the first
162. 5:22chapter of every course is for free, so
163. 5:25you can start right now with the Python
164. 5:26track. And if you want to continue, you
165. 5:28can use the link on my description. I
166. 5:29can get you 25% discount. Now, let's go
167. 5:32back. All right. Now, to the next
168. 5:33question, which tools and platforms do
169. 5:35companies ask for? All right. Now, to
170. 5:37the next question, which tools and
171. 5:38platforms do companies ask for? Now,
172. 5:41here we have a long list and they are
173. 5:42very close to each others. So, at the
174. 5:44top, we have the LangChain, then the
175. 5:46PyTorch. After that comes the LangGraph.
176. 5:48Again, there is no big decision for
177. 5:50this. And about the model providers,
178. 5:52here we have the two big names, OpenAI
179. 5:54and the Anthropic. ChatGPT or Claude.
180. 5:56Currently, my data says OpenAI is
181. 5:58winning with little bit percentage than
182. 6:00the Anthropic. So, the job market didn't
183. 6:02decide for you, and you can see Gemini
184. 6:04at the end. After that comes TensorFlow,
185. 6:06Llama Index, and Hugging Face. And to be
186. 6:08honest, I was little bit surprised about
187. 6:10the Hugging Face. I was expecting this
188. 6:12to be mentioned more often. Now, of
189. 6:13course, the question is, should I go and
190. 6:15learn all those stuff? Well, of course
191. 6:16no. My recommendation here, go and pick
192. 6:18one framework like the LangChain or the
193. 6:21LangGraph, because they are built for
194. 6:23agents. And about the providers, it's
195. 6:25enough to learn either GPT or Claude.
196. 6:27For me, in my everyday, I use currently
197. 6:29Claude, but I might switch. And for the
198. 6:31PyTorch and the TensorFlow, I'm going to
199. 6:33say leave them for now because they are
200. 6:35meant for training models. And as I told
201. 6:37you, it's not your main job. Now, after
202. 6:38that comes the cloud platform. Of
203. 6:40course, many companies are now running
204. 6:42on the cloud, and at the top we have the
205. 6:44AWS, then Azure, and at the end comes
206. 6:46Google. You don't have to go and learn
207. 6:48all of them. Go and pick one of them.
208. 6:50And I'm going to say follow the numbers,
209. 6:51going to be AWS. Now, we move to the
210. 6:53last group about the deployments. Of
211. 6:56course, our job will not end after you
212. 6:57build the platform. You have always to
213. 6:59learn how to deploy it, how to bring it
214. 7:01to the target environment, to the
215. 7:03production. So, the first thing that is
216. 7:05mentioned is working with the APIs. Then
217. 7:07after that comes the CI/CD and the
218. 7:09Docker. And I can tell you I've met a
219. 7:11lot of people that are learning AI
220. 7:12engineering, and they skip this. And
221. 7:14this is one of the biggest mistakes that
222. 7:16I've seen a lot of people that are doing
223. 7:17as they prepare to become an AI
224. 7:19engineer. They just learn how to build
225. 7:21things, and they totally skip the
226. 7:22deployments. This is what going to makes
227. 7:25you stand out, and this is what makes
228. 7:26you an AI engineer. At the end, you are
229. 7:28an engineer, not someone that's just
230. 7:30playing with the AI at their home. All
231. 7:32right, my friends, by looking to all
232. 7:33those skills that are expected from you
233. 7:35as an AI engineer, you understand that
234. 7:37you are a software engineer that you
235. 7:39learn how how use the language model to
236. 7:41build an AI system. And your tools going
237. 7:43to be Python and framework like the
238. 7:46LangChain in order to build a rag with
239. 7:49agents inside it. And you have to learn
240. 7:50how to deploy it. So, take it from your
241. 7:53local machine to that productive
242. 7:54environment. And it is not about
243. 7:57training models.
244. 7:58>> I do declare.
245. 8:00>> All right, my friend. So, that's was for
246. 8:02the skills, the first part. But now we
247. 8:03can speak about the other part, which is
248. 8:05more important and honest. What tasks
249. 8:07are expected from you as an AI engineer?
250. 8:10So, that means what the companies are
251. 8:11going to expect from you to do every day
252. 8:14at their projects once they hire you.
253. 8:16Now, for this I analyzed the section
254. 8:19what you will be doing. And here again
255. 8:20we have usually like a ranked list with
256. 8:23the most important at the start. And as
257. 8:24you go down going to be nice to have.
258. 8:26So, now the numbers are saying building
259. 8:28the LLM application it going to be the
260. 8:30number one task that you will be doing.
261. 8:32And of course there is no surprise. You
262. 8:34are hired as an engineer to build
263. 8:36applications. Then the number two you
264. 8:37are going to build the AI agents. Of
265. 8:40course we are not talking anymore
266. 8:42chatbots in the companies. The companies
267. 8:44want the models to take actions and
268. 8:47finish the work. They want the system
269. 8:49that you're going to build to automate
270. 8:50their processes. The third task it's
271. 8:52deploy it and run it in the cloud. This
272. 8:55is the most underrated between people.
273. 8:57Again, they learn how to build but never
274. 8:59how to deploy. So, here they expect from
275. 9:01you to deploy it to their productive
276. 9:03environment and as well to keep watching
277. 9:05it. If something breaks in the
278. 9:07production, you are the one that can go
279. 9:08and fix it. Now, to the next one is
280. 9:10going to be the hardest one, evaluating
281. 9:12the AI. Companies going to ask you to
282. 9:14measure the result with the benchmark
283. 9:17testing sets and scores. Because just
284. 9:19calling the AI models and build a system
285. 9:21it is not that hard. But the big
286. 9:23question is can we trust it? Can we
287. 9:25trust those results? So, the companies
288. 9:27are expecting from you as an AI engineer
289. 9:29to know how to do this. And if you learn
290. 9:31it, you going to really stand out the
291. 9:33crowd. Then we have many other things
292. 9:35like working across teams. And of course
293. 9:37integrating the model with the existing
294. 9:39systems, building the back end and the
295. 9:42APIs, building the rack, keep it safe
296. 9:44and compliance. But of course, those are
297. 9:47the main things that you have to learn.
298. 9:48And many other things. So, by looking at
299. 9:51this, this going to be your day. You are
300. 9:52building the AI, the software around it,
301. 9:55you deploy it and take care of it. You
302. 9:57measure whether we can trust it or not.
303. 9:59This is the whole thing what they expect
304. 10:01you to do every day at their project
305. 10:02once they hire you. All right. Now,
306. 10:04moving on to another interesting
307. 10:05question. How much do companies actually
308. 10:07pay for AI engineers? Now, just warning,
309. 10:10those numbers are only from the US job
310. 10:12market and only 30% of the ads in USA
311. 10:15mention a salary. But I still I want to
312. 10:17show you this to get some feeling about
313. 10:19it. But of course, the numbers are
314. 10:20totally different at your country. In
315. 10:22USA, the average for juniors going to be
316. 10:24150k, for seniors around 200k, and the
317. 10:28leaders 235k.
318. 10:30Again, average and only in the USA. But
319. 10:32I compared those numbers with the other
320. 10:34roles like a data engineer, and I can
321. 10:37tell you this is higher at every level
322. 10:39compared to the other roles. So, yes,
323. 10:42companies are willing to pay more for AI
324. 10:44engineers, and this is totally expected
325. 10:46to be honest, because now most of the
326. 10:47companies want to become AI-driven
327. 10:49companies, and they need AI engineers to
328. 10:51implement their strategy. So, good money
329. 10:53at AI engineering. Now, moving on to
330. 10:55another question. Do companies want a
331. 10:57degree? And are they asking for
332. 10:59certificates? Look at this. Of course,
333. 11:01same surprise like the other videos.
334. 11:03Half of the AI engineer ads are not even
335. 11:06mentioning a degree. So, it's totally
336. 11:08possible to enter this role without a
337. 11:09degree. But of course, if you get one,
338. 11:11you are just hiring your chance. Now,
339. 11:13about the certificates, not even 3% of
340. 11:15the ads are mentioning one. So, there's
341. 11:17no reason to go crazy and start
342. 11:20collecting certificates left and right.
343. 11:22And always my recommendation to you is
344. 11:24do end-to-end portfolio projects. This
345. 11:26is way more important than collecting
346. 11:28certificates, and for sure this going to
347. 11:30helps you a lot during the interview
348. 11:32process. It's always better to explain
349. 11:34your portfolio project in the interview
350. 11:36than just showing certificates.
351. 11:38>> Sir, you're awesome. Here is a plaque.
352. 11:41Here is a certificate.
353. 11:42>> All right, my friends. So now, let's
354. 11:43talk quickly about the other AI tool
355. 11:45that I promised at the start. It is very
356. 11:47close to AI engineering. We have the
357. 11:49trending forward deployed engineer. So,
358. 11:51what is that? It's like solution
359. 11:53engineer, but for AI. At companies in
360. 11:56the industry, we have big issue that the
361. 11:58AI projects are failing so hard. Yes,
362. 12:01the companies have access to amazing AI
363. 12:03tools, but they are failing to build and
364. 12:05integrate AI systems directly at their
365. 12:08tools, at their processes. So now, if
366. 12:10you go to the company, you're going to
367. 12:11find that they have hundreds of AI
368. 12:13projects that are built so fast, but as
369. 12:15well, they are dying so fast. So, many
370. 12:17of them are going to the garbage, and
371. 12:19that's all because of one reason. They
372. 12:21are failing to integrate with those
373. 12:23systems at the core systems of the
374. 12:26company. And this is, of course, a big
375. 12:27waste of money and time. And this is
376. 12:29exactly where comes the role forward
377. 12:32engineer. They have first to understand
378. 12:34the requirements from [music] the
379. 12:35customers and take them step by step
380. 12:37through the whole process until they
381. 12:39finally get integrated AI system to
382. 12:42their core tools. So, they are
383. 12:44responsible of the whole journey end to
384. 12:46end. [music] Now, in order to understand
385. 12:48the difference between it and AI
386. 12:49engineering, think about it like this. A
387. 12:51bank might go and hire an AI engineer at
388. 12:53their company in order to build an AI
389. 12:55system. So, they are directly hired at
390. 12:57the company. But now, if you have a
391. 12:58company like Databricks, a platform, a
392. 13:01vendor, they might go and hire forward
393. 13:03deployed engineer and then send him or
394. 13:06send her to the bank, to the customer of
395. 13:08the Databricks, in order to build and
396. 13:10integrate Databricks directly at the
397. 13:13bank. And they might start working, of
398. 13:14course, with the AI engineer that is
399. 13:16hired at the company. Just with one main
400. 13:18goal, to make sure that Databricks is
401. 13:20working correctly at the bank. So, that
402. 13:22means, my friends, this new role is
403. 13:24trending so much at the platform, at the
404. 13:26vendors, like Databricks, OpenAI, and
405. 13:29Anthropic. So, if you are targeting to
406. 13:31work there at those companies, then you
407. 13:33might consider forward deployed engineer
408. 13:35more than AI engineering. But, anyway, I
409. 13:38have done the study and I have compared
410. 13:39those two side-by-side. Now, by
411. 13:41comparing the skills, here's the thing,
412. 13:43there is no big difference between them
413. 13:45at all. Yes, you have to learn Python,
414. 13:47LLMs, agent track, and so on. So, that
415. 13:49means switching from AI engineer to
416. 13:51forward deployed engineer does not
417. 13:53require any new skills. But, once I
418. 13:55start comparing the tasks, you're going
419. 13:57to see exactly the difference between
420. 13:59them. As a forward deployed engineer,
421. 14:01you build the system at the customer
422. 14:03side, inside their system, and you're
423. 14:05going to stay there until people
424. 14:06actually use it. And more than 80% of
425. 14:09those ads have a task about the
426. 14:11customers own systems. And half of them
427. 14:14talk about as well traveling because you
428. 14:16have to go to the customer as well. And
429. 14:18as well, I found many ads that they want
430. 14:20people who were before solution
431. 14:22engineers. So, my friends, both have the
432. 14:24same skills, you are building the same
433. 14:26thing at the end, but they have
434. 14:27different life. So, if you want to build
435. 14:29applications directly at the companies,
436. 14:31then you have to become an AI engineer.
437. 14:33But, if you want to go to the customer
438. 14:35and work at the platform themself, at
439. 14:37the vendors, then the forward deployed
440. 14:39engineer going to be the one for you.
441. 14:41All right, my friends, so with this we
442. 14:42have covered the analysis from the ads,
443. 14:44and now we have to go and put everything
444. 14:45together into a road map for AI
445. 14:47engineering. The first thing that you
446. 14:49have to learn, there is no discussion
447. 14:51about this, is Python. My friends, you
448. 14:53have to master it, not only learning the
449. 14:56basics, you have really to go into
450. 14:57advanced topics. Then, after that, I'm
451. 14:59going to say go and learn about APIs
452. 15:01because AI applications at the end is a
453. 15:03service. Other softwares from the
454. 15:05companies going to go and call it, and
455. 15:07your system has to react and answer
456. 15:09back. Then, the next one is Git because
457. 15:11you are coding, and then CI/CD for the
458. 15:13deployments. So, for no AI, that's why
459. 15:15the second part, it's all about AI.
460. 15:17First, you have to understand the LLMs,
461. 15:20what you can do with the models, what
462. 15:22you cannot do, how to control the
463. 15:23behavior of the LLMs. Then after that,
464. 15:26learn about AI agents, where the model
465. 15:28calls your tools, follow your plan, and
466. 15:31finish the task. Then the next step,
467. 15:32it's all about learning RAG. You can
468. 15:34learn how to build a vector database and
469. 15:36feed it with data, so that the model
470. 15:38answers from your company's document
471. 15:40instead of just making things up. And
472. 15:42the last one, please don't skip this, is
473. 15:44to learn how to evaluate the AI. You
474. 15:46have to learn how to prove that the AI
475. 15:48system that you are building can be
476. 15:50trusted. So, that's was about AI. Now,
477. 15:52moving on to the third one, it's all
478. 15:53about tools and platforms. Here you have
479. 15:55to make three decisions because you have
480. 15:57to pick one for each. So, about the
481. 15:59platforms, as I said before, either go
482. 16:01with the LangChain or the LangGraph. For
483. 16:03the AI provider, I'm going to say maybe
484. 16:05go with big clouds. At least I prefer
485. 16:08using it. And then the last one, you
486. 16:09have to learn about cloud platform, and
487. 16:11here the direction is more tending to
488. 16:13AWS. But of course, always check the
489. 16:15market at your country because here in
490. 16:17Europe, we tend to use Azure more than
491. 16:19AWS. Now, to the last phase, this is the
492. 16:21most important one. You are going to
493. 16:23build end-to-end portfolio projects and
494. 16:26try to make it like a real use case, not
495. 16:29a demo, not you are building a chatbot
496. 16:31or like just uploading a PDF and then
497. 16:34the AI going to answer a few questions.
498. 16:35No, you have to build a full AI system.
499. 16:37So, you have to be building a RAG,
500. 16:39building agents, so it answers from the
501. 16:42documents and as well it takes action.
502. 16:44And you are recommended to use the fast
503. 16:45API. Third, you have to evaluate it, so
504. 16:48you write like around 15 questions where
505. 16:51you know already the answer and run
506. 16:53them. Then you go and change the prompt
507. 16:55and run it again in order to compare the
508. 16:56scores. So, that you show that you know
509. 16:59how to evaluate your AI system. And then
510. 17:01the last step, you have to learn how to
511. 17:03keep it safe and fast. So, you should
512. 17:05teach it how to refuse, it should never
513. 17:07like leak informations. So, you have to
514. 17:09learn how to make your own laws and
515. 17:11rules inside the system. And the last
516. 17:13step, of course, in your project, you
517. 17:14have to show that you can deploy it. A
518. 17:16container, a pipeline that test every
519. 17:19change on the cloud. You just need a
520. 17:20container, a pipeline that test every
521. 17:23changes that you do, and then one cloud,
522. 17:25and then maybe leave it running for a
523. 17:26week or something in order to check the
524. 17:28logs and see if anything goes wrong
525. 17:30during this week. And that you can show
526. 17:32that you can deploy things and as well
527. 17:35be responsible for the system. Again, my
528. 17:37friends, this going to be the way to
529. 17:38stand out the crowd to show that you
530. 17:40have built something end-to-end. So,
531. 17:42don't rush it, treat it as your baby,
532. 17:44and once you get invited to an
533. 17:46interview, talk about this portfolio
534. 17:48project. So, this is the road map to
535. 17:50become an AI engineer, completely based
536. 17:52on facts and the job market. All right,
537. 17:54my friends, now we come to my honest
538. 17:55opinion about this role, the AI
539. 17:57engineering. I'm going to say if you
540. 17:58didn't decide yet between data,
541. 18:00software, and AI engineering, this one
542. 18:02is really promising. AI engineering is
543. 18:04one of the highest paid role in between
544. 18:06them, and at the same time you don't
545. 18:08need like PhD or to be genius in math.
546. 18:11You just need to be coder and builder.
547. 18:13But, the hardest part of that the tools
548. 18:15and expectations from companies change
549. 18:17very fast for AI engineers. So, that
550. 18:19means you have the risk of learning
551. 18:21something that get outdated very fast.
552. 18:23But, I'm going to do my best to keep you
553. 18:24updated.
554. 18:25>> Updates are ready. I should update.
555. 18:27>> But, now in the other side, if you are
556. 18:29already a data engineer or software
557. 18:31engineer, expanding to AI engineering
558. 18:33might be the best expansion that you do
559. 18:35in your career. You are already an
560. 18:37engineer, you just need to learn some AI
561. 18:38topics and you are good to go. And
562. 18:40anyway, we data engineers in the
563. 18:41industry, more than 30% of our tasks
564. 18:44including AI topics. All right, my
565. 18:46friends, so that's what for AI
566. 18:47engineering based on the job market. In
567. 18:50the next video, I'm going to show you
568. 18:51the numbers for data scientists. If you
569. 18:53like this type of content and you would
570. 18:55like to support that channel, then
571. 18:56subscribe, like, and comment in order to
572. 18:58reach nice people like you. If you're
573. 18:59still here, thank you so much for
574. 19:00watching and I'm going to see you in the
575. 19:01next video. Bye-bye.
576. 19:03>> I'll catch you on the flippity flip.
577. 19:04Bye.