Complete Roadmap to be an AI/ML Engineer | 2024 | Artificial Intelligence & Machine Learning Career

I have seen a lot of startups offer 10 to 12 lakhs starting package knowledge skills or expertise develop to come into this field psychic learn is there and if you're doing deep learning then there is spy TCH and T of flow would you suggest like people from any specific background to come into this field diverse backgrounds just because so like a pharmacist can come into this field is interested one is going to be the Diplomatic answer and one is going to be the straightforward answer job opportun hi everyone my name is for this I have invited a very special guest on my channel his name is gopala and he's working as an artificial intelligence 3 in Google and in experiencei he's going to give us a lot of insights that how you can build your career in the field of artificial intelligence and machine learning so podcast interesting so do watch till the end and share your feedback so hi gopala welcome to my channel can you briefly introduce yourself to my audience hi shangi uh I'm gopala I work as a AI engineer at Google um I've been working here in the past two years last those hour have been working at Google little more than two years um and almost all of my work evolves around engineering and stuff like that so I've had the past experience is in machine learning as well so machine learning algorithms artificial intelligence I've have been you know working with them day in and day out possibly from 2017 great great great so uh gopala a lot of people get confused in the terms between Ai and ml so I want to know that what is the difference between artificial intelligence and machine learning that's a great question um often I also used to get confused back in the day uh it's very natural to get confused with these terms um so so when we talk about machine learning it's a discipline of ai ai is the biggest umbrella you have different things like deep learning machine learning reinforcement learning lot of different things uh but that's you know that's all the technical jargon you know this is all things that you want to tell someone this is how you sort of tell people that you know stuff but if you really want to understand what this the difference between Ai and machine learning it's the simple thing that machine learning often is the way of making machines learn by giving them examples right by showing them Trends and data and making the machines that is machine learning and AI is the broad aspect which touchs things which may or may not include data right because there are algorithms it can work you know without data so like search algorithms and all those kind of things so AI is the broader field and machine learning is within that field okay [Music] so machine learning means making the machine learn something artifici right and artificial intelligence is also providing that intelligence to the machine artificially so that's why it's quite confusing but thanks for clarifying so back then but back then when you started buzz buzz word n so how did you start it to come into this field how did you decide to come into this field uh yeah I mean interesting stories uh back in the day uh you are right that time um you know people used to be really uh looking forward for typical big big neural network kind of a models back in 2017 2018 um I started because I was really interested in computer vision uh I had done a lot of different projects for computer vision and uh this was something which I always like doing algorithms is something which I really liked and one find day I stumbled across uh you know YouTube videos of know technical technical channels and all those and it was not new as in the term has been around for like probably two to three decades but time it was new amongst the researchers they were doing a lot of things with neural neural networks and how do you implement it yourself so I started learning about neural networks the tzer flow uh was released a couple of years back it was quite new so I started working on tenz Flow it it looked very complex to me to this date I find it very complex that's the honest answer but uh that time yeah that time the uh you know the student in me thought comple so I started working on that eventually built my skills from tlow and then different different things came up and that's how I sort of entered into the world of AI interesting interesting so someone who is going to start his or her career in this field so what according to you should be the fundamental skills that a person should acquire knowled or exper Dev to come into this field um shivangi I personally feel ke you know no matter what type of AI or machine learning algorithm you want to go ahead with you should always have very good uh grasp of the fundamentals uh you know that fundamentals could be fundamentals of AI fundamentals of machine learning understanding how statistics work those things are very important uh so if somebody asks me I would definitely says Concepts and you should be so uh you know so confident in them that even if you know even if you are just Just Awaken aoke from your sleep you know you know all those things those statistical Concepts you know these things and the reason why I say this is because technology Ai and machine learning it's one of those fields right now which is evolving very fast right so the one thing that is always going to be constant is the fundamental constants H understandable so uh would you suggest like people from any specific background come to come into this field yeah diverse backgroundsi um I would say that there's no restriction on background in fact I have personally seen in my career in my current organization in my past organizations uh anyone who has a you know knack for understanding data and working with big data who wants to build algorithms can come you know some people have their experties in system design know algorith Des but AI is one such field which accepts all if you just have that nag and you want to really do something in aii you should definitely try to break it so like a pharmacist can come into this field if he's interested absolutely absolutely why not okay so someone and he want to do a self study what road map would you suggest step by step can we discuss that question what should be the you know step by step uh process understanding May there is no I mean there's no fixed process but what I would definitely recommend anyone who is doing or entering into machine learning is that first of all you learn the fundamentals basics of math statistics understanding definely you should have an understanding of statistics and maths after that I would definitely recommend to do course of there's a Andrew NG course it's a very nice course it's a pretty old one at least I used to do it back in the uh deep learning. course fundamentals of you know deep learning and all those things Basics build and it gives you Advanced logic but even before that I would say up uh there's a very nice and free course by Google uh it's called Google's machine learning crash course okay very nicely curated touches on all the topics that are required touches on most important topics and you know mostly misunderstood topics uh by a lot of people and tries to explain it in a very very simple to Bas Statics you can still do that and you'll feel very confident after that course because deep and and it touches a lot of different algorithms couple of don't try to do that first so once the course part is done courses duration generally uh it is all of them selfed but if you a beginner it might take anywhere from 2 to 3 months to complete these like I would definitely in see how you work see you know how that algorithm fits in the bigger picture of things existing systems building something from scratch may not be everyone's you know cup of te because there are a lot of things in this role and not everyone is a researcher resarch interest about these newer things uh even if the paper is already implemented you try to implement it in your way and see what is the difference that way you will get to know where you stand as compared to in the world and once your project is done you can tailor your resum proper resum because you can mention now you can mention the courses or whatever you have learned from the projects and then like you can search for a good job so um is there any specific programming language that we have to learn to come into this field um actually programming languages which people would recommend but only if you are interested in technical roles technical roles you can learn python python is a good starting point uh data analytics time people used to use r as well R language uh that is Al a good starting point but mostly python is a good starting point and python is easy to learn as well so if you're language learn you can directly start trying to solve problems which is an advantage in my eyes right I am personally not a very firm believer of you know trying to learn languages end to end and then implementing stuff I am more of a we'll figure it out while we are building it so python is the perfect language for that in my opinion then you have to learn some programming languages like python so definitely I would say like I've seen a lot of people being progr manager manag product managers lot of different roles are there definitely um you know analyst type a first comp what are the implications of that and even uh you know understanding how uh different uh different AI customers or different customers require AI for their use cases you know business problem solving that is also very important right we need people technical about the AI and also the actual business problems because without business problems AI is just another tool right right right so um as you talked about the programming languages just python or Ai and machine learning generally if you're going into a technical role framewor tools so we generally have very little tools that we work with Frameworks knowing Frameworks is much more beneficial Frameworks but tools you can use Frameworks right and Frameworks like um psychic learn is there and if you're doing deep learning then there is spy torch and tensor flow uh if you're doing Advanced uh you know research and stuff like that then Jax is a framework JX Jax you can read about that learn more about that you know lot of papers are now implemented in Jacks um so these kind of these are couple of Frameworks which are very much used in the industry and it is good to know these Frameworks if you're trying to go for a uh you know job in machine learning or deep learning or artificial intelligence in general so these Frameworks are definitely useful than M for amazing amazing so would you suggest some platforms to practice the Frameworks or algorithms yeah absolutely uh the best platform where everyone I know machine learning they all start from the same place it's Google collab uh you know Google collab gives you the perfect environment you can it's free of cost you can pay for upgrade but actually fre of they also give you a free of cost GPU G training if you have a powerful Amazon web services they all provide uh free credits uh and uh you know you just need to sign up with your card credit card debit card around $300 $400 depending on the cloud provider to practice on that but Google collab is a better option in my opinion take care soting you said to come into this field a good understanding of mathematics so and then you said backgr can come into this field so someone let's says background maths or statistics strong but he's good at programming languages can that person try to come in this that's a good question in my opinion that person can also come uh who doesn't have a very strong maths and uh statistics skills and they can if they know programming if they know programming that means they already have good analytical skills problem solving so they can eventually come and learn the maths and stats right it is I wouldn't say it is very difficult if you have good analytics there are very there are good YouTube channels there are good books uh good good courses on this available on course era and then there's couple of them available on plural site that person can read about statistics and basics of machine learning and they can start building that intuition because just like programming understand understand so learning process is there but definitely it's not impossible H either so for you either you can take step one and step two or step two and then step one it's it's the same you are eventually going to reach at the same place so either you can start from logic building and programming and then learning statistics or mathematics or you can take the other route so both ways it works so like you've been working in this field from a couple of years so is there some misconception or like some myths for coming into this field like I have heard a lot of baski in as such salary ranges diversity in so what are those misconceptions and myths that is there in the industry right now uh I think these are all you right these are all things that people say these are all myths uh and this is a myth right that AI will take up a lot of jobs I look at it in a different way I feel that a both jobs burning things jobs that don't exist now or the jobs that don't exist let's say 5 years back will now start existing uh other myths are like uh you know that AI is very hard feel to get into it's very tough to crack it I don't think so I feel that uh it's just like any other field right if you spend some time up time spend effort and you read about things you get to know the more fundamentals eventually I be you know you'll become a good uh developer you'll become a good AI engineer happy if you you can make yourself a good career in this and uh I mean I don't think that the you know personal opinion again I don't feel that the uh uh you know the fact salary is little different or maybe lower that's I think no longer true I mean we all know how much uh generative AI has caused demand for you know engineering especially in the field of AI so it's it's an Ever growing field and uh it's it's one field where you know there the landscape opportunties is very big because as compared to other existing program uh I would say Shang you would not believe but most more startups in India are actually doing uh Innovative Works in AI in machine learning they are putting a lot of money I have friends and I have people that I interact with start adoption of AI and maybe even generative AI in their teex stack a very large extent rights it's definitely agree if you get in the right time and start working in the right teams um so this brings me to another question like how can someone um gain some experience practical experience or interships or build some good projects to build a successful career in this field uh good question shivangi uh I I feel ke uh good experience starts from good knowledge uh so the first and foremost thing should be to understand the concepts very well and then start building in projects right there are a lot of Open Source GitHub reposit notur and that's a good thing if the more you interact because it's a new field right the more you interact the more you learn don't think that what you are doing is small what you are doing is also pretty big it could be the next big thing so that is where you start from and then eventually you know fundamentals and programming stats Ma you know to a certain level you're confident and that a I wouldn't call it res but portfol portfolio right and it gives you confidence also tomorrow if you sit in an interview you know a lot of Concepts Basics machine learning de learning those things become the basics and that is what you want you will get a lot of internship opportunities you'll be able to was the interview to H so how can we get the internship opportunities once that portfolio is built right what ways shall we follow to get the internship opportunities uh I would say that if you are if if you you're a student trying to look for it I would say definitely build a good and strong profile on LinkedIn uh uh you know good profile on LinkedIn B and try to you know interact with people who are working in the same field as you or whom you want to join in like you want to be in the same field as they are follow inter what all things they post on interact with those people and try to build genuine relations because when you are a fresher when you are a new person in the industry uh you don't have a lot of accomplishments anyone for that matter of right right so the only way you are going to get in is when you show the work that you have done is really good through your projects that you have done and through interacting with the right people so right people and right projects will give you all the opportunity so LinkedIn is one great flat platform along with LinkedIn or recruiters are having their eyes on to catch students especially freshers yeah uh I I feel that LinkedIn like you said LinkedIn is a good good platform apart from that I a lot of times you know colleges and you know a lot of these big mnc's have a lot of hackathons uh right I don't remember recollect the name but a lot of MC almost all Hacker have a know track you know participate once twice I've seen teams who did not even reach let's say the finals they reach the semifinal but even they were called for you know interview for an internship you make your presence F and try to build a good U what say team with which you can you know take part in these competition that will give you that additional uh you know and then you are able to do more with less time so I would say that is also an important thing recruiters as sear for freshers to be honest uh actively look out you have to look out actively for recruiters so that happens when you go to these events when you go to these offices that happens that time have you since you are at a good position in Google have you ever taken interviews also yeah I have Tak care interviews okay what what do you look in a good candidate it's in general shangi it's a mix of uh knowing technical having good technical skills definitely uh but also good soft skills and knowing how to solve problems is also important and I know when I say this knowing how to solve problems is very difficult from is very difficult and different than compared to having uh you know good technical skills uh technical skills can be often times is like it can be bookish knowledge as well right to when we try to interview we as in most googlers when they try to interview they try to look for a lot of parameters it's not just one thing we try to look for how good the technical fit is how good the person is in general uh how do they interact can they solve problems um you know how often how fast can they think and also um you know if there is something which they're doing wrong they make a mistake or something like that can they correct it themselves or not so it's a lot of combinations of things that we you know actively look for in people hhm understood okay so in a generic SD role we get some DSA questions couple of couple couple of DSA rounds then there is this behavioral round then HR round and done we get the call letter if we are selected how is this what sort of questions are asked in the interviews for this role uh so since AI is so new I've had different experiences in different places okay but there are some couple of common points for example DSA is still asked if you are a fresher even if you're not a fresher DSA is required because it helps show problem solve DSA is there H um and then apart from DSA there often happens rounds data science questions fundamentals of data science theoretical but also sometimes practical coding based you have to write small pieces of code let's say pipeline code you know processing code all those kind of things have there and sometimes the interviews that I have been in uh you know sometimes the questions are also related to machine learning and deep learning sometimes machine learning fundamentals but also system design so how do you scale this thing for let's say 100 people to 1,000 people to 1 million people what will you do different and then more deeper Concepts come in like you know concept drift and data drift Concepts respons Rel so those kind of questions come in when you go into deep technical uh rounds uh machine learning and deep learning Concepts so I would say it's very different but DSA is important databases are important database I forgot to mention and been inart from answering theoretical questionsq queries and machine learning and deep learning is a must that's the fundament but if someone is specifically applying for AI so whatever you have told like understanding of DB or computer fundamentals per se then ml deep learning NLP would it be too much um you're right it's a lot of things I understand that and uh I know it's not that possible and feasible to know everything especially for the college going students someone is not from that background yeah uh yeah so I would say generally people they break in or they enter this field by doing one thing for example if you're good in datab noq Analytics then go to let's say dashboarding power Le and then learn to come into machine learning and deep learning I took some time then I went to machine learning then I went to a systems then I went to system design then I went to uh so on and so forth I went to let's say databases anal and all those things would eventually learned things and I'm still learning a lot of things H it's a process what continuous learning is throughout your career throughout your life I would say absolutely yeah can you suggest some platforms to practice for interviews not everyone has people to like have mock interviews friends sometimes colleagues are also not there to practice mock interviews so any platform uh plat platform let's see so uh I would say for machine learning or AI UHS lead code is the good platform to practice your DSA for machine learning and AI I would say you should uh take quizzes there are a lot of comp participate and try to I would Sayers machine learning and you may not get a lot of people and right now there's not a lot of I'm not sure if there are a lot of platforms that cater to that's why that question came in my mind right right so there is no one fixed process what I would say is that and apart from that you know these platforms like you know platforms like uh there's a lot of good subreddits on Reddit as well which will help you give you quiz questions and all those things but more importantly in my opinion because machine learning is a very is a very analytical and applied field when I say applied is that that means right more than checking your knowledge which is just um you know a b c what will be checked is that how do you approach a problem how do you solve a problem what are the thoughts that go in your head so then you solve a lot of problems so take part in a lot of competition go through a lot of GitHub repos use them develop lot of stuff Clarity confidence automatically for those ints makes sense makes sense now coming to the most asked questions is it at with the SDS what are the ranges that we get so I will give you two answers to this question one is going to be the Diplomatic answer and one is going to be the straightforward answer okay uh the Diplomatic answer is that the salary ranges will be as high as you are as you as high as good of an engineer you are okay right if you are a really good SD it will be high if you are a really good AI engineer it will also be high but realistically speaking I would say in my experience I have seen the growth of AI engineers and AI roles in general not just engineering salary growth has been much much higher the rate of growth has been much much higher over the past 2 years and the range I mean the range goes I mean different organization different but I have seen a lot of startups offer 10 to 12 lakhs starting package uh again these are startups packages so if you go to tech companies big tech companies and just in a couple of few more years you can you know you can have good increments I've heard a lot of my friends got like uh 20 to 25% increments you know doing these kind of machine learning roles and uh MH you know even like deep learning scientist all those things you definitely have a lot of potential here I would say because project there's a lot of impa ma learning get a lot ofil right right right um do you remember some common companies that always keep hiring for these roles startups or any service based companies any product based companies like I've heard mintra hires a lot in this uh misho hirers dat scientists um then there are also these are product companies then there are uh you know analytics companies service based sometimes like I think fractal is one that there's tiger analytics uh these are like good companies and they give you very good uh uh they give you a really good uh exposure uh another one is quantify that's where I also work uh this is like these are like good companies to start your career with and definitely you'll learn a lot of things these companies are good options H so so companies mainly dealing with a lot of data are potentially hiring for these c yeah understood huh basically H but the thing is that India because of our population there's anyways a lot of data in most of the startups have a lot of data by default so every company eventually will become a data company so U you have a lot of experience in diverse ways diverse projects so did you face any challenges while working on these projects and would you like to give idea on what type of project should you work on absolutely sh uh challenges I faced a lot of men this because mostly machine learning is not very clear that is not very well defined uh Tessa generally you know you have SD projects you know the end State you have a good idea about what you want to achieve and how will you achieve it in AI projects there's a lot of ambiguity the last is going to work so most of the problems revolve around data or model mostly it's around data data quality quantity sufficient all those things and even after you build a model data limited challenges frequently and I have asked this lot of my colleagues as well everyone faces these kind of challenges uh especially with machine learning projects AI projects it's a common Trend and this is something that uh with these kind of you know ambiguous requirements ambiguous situations yes hard time deterministic solution and you need to be okay with it it's your job to find out uh you know a solution Clarity that's a part of your job so that is the challenge and yeah across the industries I worked across different Industries worked in I worked across Finance Industries gaming industry uh media Healthcare Industries so challenges remain the same more or lessance indes because Finance Industries uh don't get a lot of access to the data and because of lot of you know regulations and rules compliance right right you have to be extra careful with the data and with media companies it's the data Fidelity is a little difficult to maintain right because in media companies it's like people it's actual human people C and you have to deal with them so so those kind of challenges are always there I would say the challenges are more or less common but uh they keep changing and as you start working experience you start to deal with the challenges in a more efficient way and train it will be trained in a better way and it will give a better output better optimized solution we could say 90% 91% 92% accuracy will improve right that's how it works right so um is it going to get used somewhere in the projects the DSA Ah that's a good question I see if you ask me DS it is going to used in some project maybe in some but not in all okay that is the right answer um but that is how it works unfortunately or fortunately that's how it works DSS problem problem solving so when you actually are facing a lot of know difficult and complex problems it's not that difficult to figure out but I would say that it is not just because true true true you get to know an approach to solve a problem or to break down problem into smaller chunks so that you can solve them better so that's how DSA works that's true so coming to the end of the podcast any advice you would like to give to the freshers who are preparing to come into this field or your tips uh I'll give some advices tips so I don't think I'm that qualified to give tips but the advices that I'll give is that uh you know this field isn't as difficult as people think it is difficult just work hard and get to the bottom of problems right if you want to really leave an impact be problem just because at least at the beginning of your career root CA understand how is it being solv because so try to go into the depths of problem solving that is one thing apart from all the things that I already said you know basics of fundamentals machine learning de learning but apart from that try to always go to the deepest you know most root level understanding of why this problem is happening or how do I solve this problem and build intuition on top of it build intuition on algorithms understand algorithms not to solve you know not to solve DSA questions but actually understand algorithms to solve problems once you transition you move from that position of problem solving and you know how to do it that is going to be the time when you will actually see a very big benefit coming from all this reading that you have and keep reading uh keep reading news about tech keep reading papers papers I know sometimes are too long but keep reading summaries of those paper abstracts of those papers say you will keep on knowing what is the new thing which is going to comeing and don't feel overwhelmed don't need to do everything at the same time things take time uh don't take a lot of pressure on yourself everyone takes time and I've seen a lot of people who are experts in their field even they fumble so it's okay don't be too hard on yourself just just start don't keep procrastinating great advice AB So yeah thank you so much goala for coming on my channel giving your time and sharing so much information it was great interacting with you thank you so much shangi for having me guys if you like the video don't forget to like share and subscribe to my channel and don't forget to hit the Bell icon so that you get notified whenever I post my new video thank you so much

Channel: Shivangi Tiwari

Views: 99,137

Keywords: Artificial intelligence, ai job opportunities, job opportunities in artificial intelligence, artificial intelligence jobs, artificial intelligence job opportunities, artificial intelligence job roles, artificial intelligence career path, career path in artificial intelligence, best ai jobs, top ai jobs, how to get a job in ai, AI Engineer, Ai engineer career path, ai job salary, ai jobs for freshers, machine learning, machine learning engineer, ai ml roadmap, ml jobs

Id: 5q4i4VObrtw

Length: 38min 40sec (2320 seconds)

Published: Sun Mar 03 2024