Just to give some people motivation; I was so bad at math. But math in data science like linear algebra and calculus is absolutely learnable. There’s so many good classes you can follow on TH-cam and online courses. Khan academy is also a good way. Just keep pushing if you really find this interesting to you.
Jhonny hey I am basketball player (athlete) who hates math. Always have. But I am doing my first ever coding course: Python for Finance on Udemy. I even downloaded the Brilliant Math and Logical thinking app to work on some math problems to start thinking and solving problems mathematically. The more time I spend the better. It’s all about time application!
When comparing job openings you can not take only the growth into account I.e. if you have 10k ML openings but 200k Software Engineering then a 200% growth similar impressive for a more mature field. Similarly you have to consider the spread in the salaries as there is usually a higher gap in the ML area. Please don’t screw data like that, especially if you work with it.
there are almost no entry level ML jobs too, so ML will usually attract more senior people who are better payed.... The girl has no idea what she is talking about.
in terms of job openings, you can't ignore the base and talk about growth. SDE has a larger demand than ML in total number, but ML is at its beginning.
Currently learning both and trying to build some simple projects so that I know which one I like and enjoy doing and will continue to pursue in the long run.
I like both react-native - node js and Machine learning ,so I decided to learn software first to get the overall picture of a software, then I will focus on machine learning
Great video, one thing you missed however is that there are far more software engineering jobs out there than ML jobs, I would say at least 10-1, its a massive difference. This is a big factor in which career could be viable for someone. Also, ML is not going to automate most of the software engineering work anytime soon. 90% of an engineers time is spent on system design/architecture, simplifying requirements, and reaching consensus, maybe 10-20% is coding, most of software engineering work is not going to get automated in our lifetime.
Excellent video! It really resonates with me since i am currently stuck in the exact same position. I did my CS bachelors with focus on software engineering but am now in a masters program focused on data analytics and AI.. I still haven't made a decision yet but my personal main decision points are: SWE: Higher demand / it feels like only bigger tech companies are doing ML/DS while you can work as a software engineer even at smaller companies ML/AI: steeper career path in the long run, since you can (if you excel at it) become one of a few data scientists in a big company, instead of one of a dozen Software Engineers and you also have the option of pursuing a phd (now you can also do a phd in SWE but i feel like that wouldn't benefit you nearly as much) It's a really hard decision to make but i am leaning towards the machine learning path right now, partially because i feel like coding projects could be more suitable as a hobby and this way, one does not have to give up neither of them :)
It is true that the demand for software engineers is larger now. It is also very evident that the future job market will revolve around data-related jobs. In my opinion many software engineers will actually do data engineering projects. However, if you enjoy mathematics, and you contemplate the possiblity of doing a PhD, then machine learning can potentially be a more prosperous career for you indeed :)
I'm about to enter college. The course is 3 + 2 years. 3 years of Computer Science + 2 years to get the Masters in Artificial intelligence... I'm quite excited to know that it will still be really relevant when I finish the college and start working in about 5 years...
The career I want, but the explicit major pathway that doesn't exist at universities :(. I'll just go down the CS route and take as many machine learning related electives as I can 🤷♂️.
best comparison video between the 2 career paths. I watch this video whenever I feel unsure about my decision in career path and it always helps me feel at peace with my choice
Great video! As an electrical and electronics engineering student currently involved in a research project using ML I specially relate to the "less code more mathematical/ analytical thinking" side of things you pointed out.
Thank you! This cleared my dilemma. I heard from some less experienced people that the scope of java developers was becoming less and less. I personally love building things. I'm looking forward to shift my career from a Marketing Officer to a Software developer in the coming year.
Both are not mutually exclusive. I developed a deep learning model, productionised it (dockerised, deployed with CICD), developed the backend API, and also help write react hooks to call it.
Although I work as an MLE (building models + productionizing them on servers), I enjoy software development part of the work more than model building mostly because of the impact I can have and how productive I can be as a software developer. Model building can be a tedious task wherein sometimes you do not have enough data and sometimes you can not achieve a desired performance. It is even worse when the model performs well on validation sets and performs poorly in the real world. Software development on the other hand gives you an instant feedback and as you mention in the video "it either works or it does not" and no inbetweens. This is just my perspective working at a startup.
Good video, but I think you should consider that "software engineers" exist at all levels of seniority, whereas the "machine learning engineer" role is almost exclusively a senior level role with virtually zero entry level positions. Anyone that takes the ML path and is entry level is most likely a data analyst and only becomes an ML engineer or data engineer after many years of experience. So the two salaries are not really comparable.
It's really a too simplified view (though I agree with many of the things you say) 1. In SE you have a lot of PHP developers and other branches that are bringing the average really down. You cannot compare the broad industry of SE to the more specific ML industry. In this case it is just simply not a good idea to compare averages. SE can still really earn a lot in a specific area (on average). 2. SE exists for a lot longer, is a lot broader, so showing a growth percentage between an emerging job and a longer existing one is really invalid. If you showed "totals" then you'd draw the conclusion SE would be better. ML Engineering is not about "coming up with algorithms". It's much more about "bringing ML models in production" in a reliable manner. Also make sure you can compare experiments and models across time. The infra around that is what ML Engineering is also about.
Thanks for your feedback! I agree that I should have provided base values rather than just growth percenatges - my bad! When it comes to ML engineers/scientists - whether their job is about bringing ML models in production or creating the ML models will depend on a particular position/company. My experience is that job titles in machine learning are pretty fluid. What I was trying to do was to contrast the role of an engineer with the role of a scientist. I hope I managed to get the general idea across! :)
I agree with your last statement about ML Engineering. She implied MLE was just "coming up with algorithms" with no "building stuff". That is incorrect and worryingly moreso in a video that may be used to convince someone to choose between SWE vs MLE. Sure there will be jobs that include a Data Scientist's job in MLE but at the end of the day, it is STILL engineering. You will be building things regardless.
machine learning engineers have to be good at math, abstract thinkers, and love theories? Have you actually worked as one? BEcause this doesn't resonate with me. In all the machine learning work I've done, I don't think I've ever had to be an "abstract thinker", what does that even mean? ML work is much more about having a intuitive bone deep understanding of how ML works and how a machine "thinks" to the point of being able to know how to approach a ML problem correctly cause there is many tools in the tool belt and we have an ocean of possibilites. Thats it, that's the real differentiator between a good ML engineer and a crap one. The math involved is not that bad at all, and you only use "math" in ML to make sure you understand what is happening behind the scenes and if that's what you expected, you're not actually doing math. Loves theories? Ok, Ill give you that one, you gotta be able to sit through and understand a good amount of nitty gritty and data philosophies.
Thank you! This video was really helpful. Especially the part about the predispositions which clarified more about the fundeamental differences between the two.
i've always heard that machine learning is about getting data-science models into production -- so like train/test splits, what are easy computational features, how do they affect your model run time, what are the run-time constraints of your model? what's your model availability constraints? i'm not sure, how i feel about everything but i'm entering a hybrid data-science/software-engineer role at my company, and obviously it's full stack or full like our application stack so dev-ops, run time analysis, deployment, creating unit-tests, helping transform a deterministic algorithm into a probabilistic one i think at a certain point role names are meant to try to "describe you" less than application to a certain role, just always keep learning and pushing yourself. try to focus some attention on what you find interesting, and you'll end up excelling in that area and keep getting more challenges towards that path
I believe that the first comparison is not fair as software engineering is way more established than machine learning so using percentages seems unfair/inaccurate. Other than that a good video as always Waiting to hear your opinion on ML for CFD
You're definitely right, software engineering is more prevalent, also in smaller cities where earnings are lower and drive figures down. Averages are not the most useful statistics in general! I'll be having a look at CFD over the weekend, chat to you soon! :)
These salaries look like junior salaries especially in UK! Anything below 50k is what I would expect student or junior dev to earn in their first year!
great content as always Karolina you nailed it! in my first two years at college, I knew software engineering is not what I want I started studying Machine Learning and Data Science all by myself last year I started my own business after only one year of studying! now I am studying public relations and media because I think if I have both with Data Science I can get even more out of it! one day software engineers will be replaced by AI ;)
That's amazing - to set up a business in a completely new field so fast is really admirable! How do you intend to combine public relations and data science knowledge?
@@karolinasowinska well, after what Cambridge Analytica did, and I studied about what they did, I was thinking why not doing quite the opposite! instead of changing public's behaviors and controlling what they think, why not focusing on how to save and recover reputation for governments, companies, and even celebrities. so as we know public relations is mostly about saving and recovering reputation and showing the face that we truly are and want to show, and with the help of data science and machine learning, we can collect the data we need, make NLP tools, analyze and visualize everything easier, and also combine them with computer vision. Netflix personalizes thumbnails, in our case, we can generate or at least choose templates and layout designs. it is not only the text that we can collect, it is also images, with flag classification for governments and parties, classifying sarcastic images etc...
@@keyo3945 That sounds really amazing! I'm always of the opinion that the best things happen when you combine knowledge from two supposedly unrelated areas to come up with unique insights/solutions. I see a lot of potential in what you say - good luck with that! :)
@@karolinasowinska thanks a lot 🤗, that is why I love data science, math+programming+whatever you want(music, agriculture, healthcare, media, economy, public relations...) this is my favorite channel now so keep up the great work!(I am always here, and try to be the first to watch your new videos)
@@keyo3945 That is so incredibly nice of you, I really, really appreciate your support! Thank you, and I'll try not to disappoint with the future content! :)
I think for your comparsion the better term would be data science, not machine learning. In my view data science is moving away from machine learning and focusing more on stats and advanced analytics. Machine learning is more and more done by machine learning engineers, because you always need software engineering to productionalize models and AutoML is getting traction.
Of course mathematically it can be! When I said that, I wasnt sure if you can have a "never ending learning curve" in English, because usually it is a "steep learning curve", and so my brain hesitated for a second!
This was very informative. I'd suggest everybody not to consider the average salary so important. Salaries don't follow a normal distribution and, therefore, the average is not very useful. For someone trying to compare salaries It's much more interesting to see entry jobs pay checks on your region and check how many job opportunities there are. In where I live, software development usually pays more than machine learning engineering but once you become a senior this trend reverses.
Hey Karolina Good video, can I ask a few questions please? well more about asking for advice, I really love math , I do like algebra and probably some statistics, planning to go to machine learning but I don't like coming up with the most efficient algorithm all the time. I like software engineering too and I love building stuff but it feels like what I learned from data structures and algorithm is kinda feel wasted? again idk why I am like this but i don't like toying around with data not to the point it's too complicated, so what should I pick if that's the case?
The problem is that I like both of them equally. I am creative, I am good at math, I love theories, and I am in love of coding stuff. I know Python. And I have a choice : Go for Python ML or Go for Java SWE. And I am totally confused. This video helped me, but I still can’t choose...
I would like to clarify some of your facts. That's not how statistician looks on the data. That's data misinterpretation. When you provide percentage as value, you need to know from what percentage it is, else you can't really compare it. The base from the indeed report is that there was a rise from ~52 millions of ML job postings to 179 millions (344%). That means *127 mil new ML job postings*. In the Software engineering it was from 401 millions of job postings to 828 millions (206%). That means *400 mil. new job postings* in SE. Seems like scam to me, a lot more job SE job postings. You forgot to mention the new machine learning technologies like AutoML which will probably replace a lot of not specialised (prob non phd) ML engineers as well as SE engineers. The simplification of using the ml models is real, look at GPT-3. Developers will probably be using it even though they don't have any big knowledge in math. Transfer learning (fine tuning) of existing models will also be simplified in the future. Other than that, some nice points were shown
Fair point - I should have provided the base values, instead of just focusing on the trends - my bad! I'm aware of GPT-3's implications and even discussed them in my latest video! :) And you can expect a video on AutoML as well, I've been meaning to do it for a while. Thanks for your feedback, that's really much appreciated, it helps me to be more relevant and accurate!
The inference of salary comparison is incorrect. ML roles typically require a masters and PhD degree, whereas you'll find a lot of undergraduates in Software Engineering.
This is a very educational video, delivered in an entertaining matter. Keep up the awesome content. In my opinion, I wouldn’t be able to do both. It makes why someone would have to pick one or the other to achieve higher success.
I actually think the responsibilities of a Software Eng. and ML Eng. are not so different. Most of the time, I see ML engineers simply handling how the core machine learning algorithms are "productionized" (deployed, integrated with the application layer, etc.) so there's not necessarily a massive need to understand all the statistics/linear algebra behind the model but just how to optimize it for use with cloud/edge/hardware/iot applications. I think Machine Learning Scientists/Data Scientists are normally the ones who hold PhDs or at least masters, which are prototyping and even creating new models from scratch. Also Software Eng. deals with plenty of ambiguity as well, depending on your domain. A web dev, maybe not so much... but if you're a software engineer working with mathematical optimization or higher order algorithms, or algorithms that are the NP-hard+ space where you need to have a solver running and ingesting a massive amount of data (think large LP/QP chained model) then things begin to get nearly or just as ambiguous as the ML modeling area.
Using the definition that you're referring to, in my video I'm actually making a distinction between Machine Learning Scientists vs Software Engineers :)
I finished my bachelor as a Software Engineer and worked for 2 years in web and now I am doing a master's in Ai and I was wondering if those two can be combined. I totally agree that maths makes me struggle but as HeliTom said if you put effort you can make it. So what's your opinion about my question do you think someone like me with experience in business web apps and a degree in AI-Data Science could combine these two for innovative solutions
I'm really confused... I don't know what to do! At the moment, I'm studying at a SE University because I've always loved coding and I've always had a crush on the world of technology since the age of 10. However, when I discovered something about ML I was so fascinated... the fact that I could teach to something that doesn't have its own mind to actually start thinking and doing stuff was making me crazy and I thought it was the right choice for me! Here's the point : I hate theory. I'm a very practical person, I love coding and I love coding stuff. I'm the kind of person that wants to finish asap a manual on a programming language because I want to put my hands on the code. I love the aspect of robotics and AI. but math and theory is not for me. Is there a way to go out of this?! Hahahah
This might be the most useful/informative 12:49 i have ever spent. Thank you so much for this, you've shined a tremendous light on the path forward for me. Also i enjoy how methodically put-together your videos are. They have an appealing logical flow :0) Thank you again.
It seems that there are many machine leaning engineers that works in many different fields(computer vision, NLP, recommender system, ads ranking, fraud section in bank, etc.). Which fields hires more machine learning engineers?
do you really think there's a big difference between ml engineer and software engineer 🤔 I was thinking software engineering is a umbrella term so ml engineering is just a branch and specific that's all
I meant that being a machine learning scientist/researcher is way different than being a software engineering. Sorry if I didn't make it very clear in the video!
She is right. They are very different and one is not an umbrella term for the other. But artificial intelligence can be considered as an umbrella term for ml engineering.
I liked the video. At the end I think you are contradicting yourself a little bit. If you build lots of things as a SE you will have to drop some optimality. It happens as well when running ML models. Another thing, I don't think that the practicality of SE is smth inherent from the job, but rather something to admire and even to apply in other fields. At the same time, I think SE could raise their heads and try to think about the value or the impact of what they are doing. Cheers
I am graduating next year with a masters in comp sci and am making this decision. I only know the basics of machine learning, but on the other hand I'm very good at software engineering. Because of this I find it hard to compare how much I enjoy them both as I still find ML hard to understand. Does machine learning always feel like hard work, or do you ever get to a stage where it can feel simple and enjoyable?
hey andrew, i'm in the exact same position as you were a year ago and would love some clarity from your experience. if you don't mind helping me out a bit, please drop your instagram/snapchat/email
I think machine learning will become a part of software engineering and that too would done by A.I itself, But if they kept on outsourcing and Sass we will loose all the jobs. Most jobs that are done by scientists (who are also artists) will flourish, Machines can do the tedious work that we have to do over and over again but If we don’t have Scientists who define and design solutions to problems for them and tell them how to function, they won’t be able to do it. A.i. Is just a fancy name for a software tool that will do things that we aren’t good at, AGI might never even be possible, due to the complications in hardware, energy consumption by such a computer, the philosophy behind it and also you can only train the neural net to process a specific type of data set, It can’t process everything.
What if you love both? I'm a seasoned software engineer who absolutely appreciates maths and I do want to get into artificial intelligence. I've been looking for cross-functional sort of role. Can anyone suggest courses, ideas of jobs around that, please?
Hey karolina, I want to ask about the automation of which tech jobs in future and which not, especially in IT sector.(( PLEASE REPLY)) and make a contentful video with facts and figures + probabilities of happening of automation or not 👍👍👍
This is a very bad video. All the numbers are manipulated to make ML look good, eg. when she said ML jobs have 300% growth but SE has 200% that's because ML has 1K jobs wheras SE has 100K - absolute numbers tell a very different story. SE is not going anywhere and I can tell you for a fact that we SEs would be very happy if we need to code less in the future. There are a lot of different things to do in SE other than writing code.
Machine learning engineering is a type of software engineering. Machine learning on its own (or data science/deep learning) is more of a research job :)
Growth alone as a metric is not sufficient. There are more software engineering jobs and using growth alone disguises that there were actually much more software engineering jobs opening than machine learning.
Wow😲 first thank you for this video, i have stud computer science for three years and i need now to choose wether software engineer or ai engineer, the problem i face that i love both math and creative building and i decided also to get master degree after the college but i still confused with my choice, what do you think? , And thank you again for this video ♥️♥️
With all these *SaaS* companies there is a small but considerable amount of decrease in Software Developer job role what do you think about it let me know?
This video is like boom!!! And what you are doing is agreat job 🙏:-)..and Yes please make more videos on career in machine learning,ai,dl and please help us know which are the highest paying companies in ML,Ai,dl field.
Thankyou so much for making this video. Can I take what you said about SE and apply it to App Development as well for the most part? If not, I would highly appreciate if you make a video analysing App Development the way you did these two? Particularly interested in the "Your Predispositions" part which I feel like is most important. You did a great job simplifying it concisely. Most people are too afraid of being reductionist when they simplify things that way.
@@karolinasowinska Okay that's great! Any advice for someone who's just about to start (again) his Undergrad? Like is there anything you would do differently if you could go back?
Super useful video. I think software engineering sounds more up my street. Though, I do find AI interesting so who knows what i may choose down the line. Currently, doing a python coding bootcamp to get into the software engineering field as an umbrella (studied genetics so maybe that will be applicable for machine learning...)
Great video! That clarifies a lot of things. I just have a couple questions - does it make much difference if you have a BSc followed by an MSc as opposed to a 4 year integrated masters (MSci)? And do you think a (separate) masters in ML is a good idea?
Do you really need a masters to get a job as a machine learning engineer? I mean like now there are tonnes of online courses out there which you can do parallel with you bachelors...?
Just to give some people motivation; I was so bad at math. But math in data science like linear algebra and calculus is absolutely learnable. There’s so many good classes you can follow on TH-cam and online courses. Khan academy is also a good way. Just keep pushing if you really find this interesting to you.
Great comment, you're so right! What matters more than current math skills are your determination and your potential to learn them!
thanks bro, I really needed it. I'll start studying it again :D
Jhonny hey I am basketball player (athlete) who hates math. Always have. But I am doing my first ever coding course: Python for Finance on Udemy. I even downloaded the Brilliant Math and Logical thinking app to work on some math problems to start thinking and solving problems mathematically. The more time I spend the better. It’s all about time application!
Someone who thrives as an ML engineer should love data . Just that can determine success .
@@Beny123 I agree! Usually when someone loves something, they are naturally good at it as well.
When comparing job openings you can not take only the growth into account I.e. if you have 10k ML openings but 200k Software Engineering then a 200% growth similar impressive for a more mature field. Similarly you have to consider the spread in the salaries as there is usually a higher gap in the ML area. Please don’t screw data like that, especially if you work with it.
Damn
there are almost no entry level ML jobs too, so ML will usually attract more senior people who are better payed.... The girl has no idea what she is talking about.
@@Daniel-xaogjeyh are you a ML engineer?
in terms of job openings, you can't ignore the base and talk about growth. SDE has a larger demand than ML in total number, but ML is at its beginning.
Fair enough, my bad! I focused so much on the growth/trend aspect, that I forgot to mention the current size of the two markets.
Good point . It is mostly the case that SWE move to MLE . The other way around much less so .
Currently learning both and trying to build some simple projects so that I know which one I like and enjoy doing and will continue to pursue in the long run.
I like both react-native - node js and Machine learning ,so I decided to learn software first to get the overall picture of a software, then I will focus on machine learning
Great video, one thing you missed however is that there are far more software engineering jobs out there than ML jobs, I would say at least 10-1, its a massive difference. This is a big factor in which career could be viable for someone. Also, ML is not going to automate most of the software engineering work anytime soon. 90% of an engineers time is spent on system design/architecture, simplifying requirements, and reaching consensus, maybe 10-20% is coding, most of software engineering work is not going to get automated in our lifetime.
True, I should have included that information too! Thanks
@@karolinasowinska How u learned ML?
10-1 sure!
there are significantly less peopel wanting to get into that domain too!
I agreed with most of that until you said "in our lifetime"
I think ML engineer is more about working with model, they even code less than SE. I prefer SE with bigdata rather than only ML engineer
Interesting point. I elaborate a bit more on ML engineers/SE/data engineers in my newest video, you might find that interesting!
Excellent video! It really resonates with me since i am currently stuck in the exact same position. I did my CS bachelors with focus on software engineering but am now in a masters program focused on data analytics and AI.. I still haven't made a decision yet but my personal main decision points are:
SWE: Higher demand / it feels like only bigger tech companies are doing ML/DS while you can work as a software engineer even at smaller companies
ML/AI: steeper career path in the long run, since you can (if you excel at it) become one of a few data scientists in a big company, instead of one of a dozen Software Engineers and you also have the option of pursuing a phd (now you can also do a phd in SWE but i feel like that wouldn't benefit you nearly as much)
It's a really hard decision to make but i am leaning towards the machine learning path right now, partially because i feel like coding projects could be more suitable as a hobby and this way, one does not have to give up neither of them :)
It is true that the demand for software engineers is larger now. It is also very evident that the future job market will revolve around data-related jobs. In my opinion many software engineers will actually do data engineering projects. However, if you enjoy mathematics, and you contemplate the possiblity of doing a PhD, then machine learning can potentially be a more prosperous career for you indeed :)
I'm about to enter college. The course is 3 + 2 years. 3 years of Computer Science + 2 years to get the Masters in Artificial intelligence... I'm quite excited to know that it will still be really relevant when I finish the college and start working in about 5 years...
If you are good, you could start working before that.
I’m definitely interested in both. I only do data now but feel like there is a lot of potential for ML-SWE down the road!
I think so too! It's an exciting time to work in tech!
The career I want, but the explicit major pathway that doesn't exist at universities :(. I'll just go down the CS route and take as many machine learning related electives as I can 🤷♂️.
@@subz424 That's a perfectly valid route that will take you there :)
I believe you can do both, I mean you can be smart enough, hard working enough and creative enough to do both :)
It is definitely true that some people are gifted enough to excel in both areas:)
Even after this my choice for the 2 split almost half n half
@Mr. Kattan it’s September Adderall season…
best comparison video between the 2 career paths. I watch this video whenever I feel unsure about my decision in career path and it always helps me feel at peace with my choice
Great video! As an electrical and electronics engineering student currently involved in a research project using ML I specially relate to the "less code more mathematical/ analytical thinking" side of things you pointed out.
I'm glad that your experience aligns with what I menetioned! Good luck with the research project :)
Thank you! This cleared my dilemma. I heard from some less experienced people that the scope of java developers was becoming less and less.
I personally love building things. I'm looking forward to shift my career from a Marketing Officer to a Software developer in the coming year.
Awesome, good luck on your journey :)
Both are not mutually exclusive. I developed a deep learning model, productionised it (dockerised, deployed with CICD), developed the backend API, and also help write react hooks to call it.
Wow, can you give me more insights on your project. I write react and nodejs as well and interested in building ML models in the near future...
SE is not just creativity ,SE require a lot of puzzle solving
That's true as well!
@@karolinasowinska software is much broader than AI engineers, software is not only algorithms but also a lot of technology.
Although I work as an MLE (building models + productionizing them on servers), I enjoy software development part of the work more than model building mostly because of the impact I can have and how productive I can be as a software developer. Model building can be a tedious task wherein sometimes you do not have enough data and sometimes you can not achieve a desired performance. It is even worse when the model performs well on validation sets and performs poorly in the real world. Software development on the other hand gives you an instant feedback and as you mention in the video "it either works or it does not" and no inbetweens. This is just my perspective working at a startup.
That's exactly how I feel about it!
Good video, but I think you should consider that "software engineers" exist at all levels of seniority, whereas the "machine learning engineer" role is almost exclusively a senior level role with virtually zero entry level positions. Anyone that takes the ML path and is entry level is most likely a data analyst and only becomes an ML engineer or data engineer after many years of experience. So the two salaries are not really comparable.
You're totally right
It's really a too simplified view (though I agree with many of the things you say)
1. In SE you have a lot of PHP developers and other branches that are bringing the average really down. You cannot compare the broad industry of SE to the more specific ML industry. In this case it is just simply not a good idea to compare averages. SE can still really earn a lot in a specific area (on average).
2. SE exists for a lot longer, is a lot broader, so showing a growth percentage between an emerging job and a longer existing one is really invalid. If you showed "totals" then you'd draw the conclusion SE would be better.
ML Engineering is not about "coming up with algorithms". It's much more about "bringing ML models in production" in a reliable manner. Also make sure you can compare experiments and models across time. The infra around that is what ML Engineering is also about.
Thanks for your feedback! I agree that I should have provided base values rather than just growth percenatges - my bad! When it comes to ML engineers/scientists - whether their job is about bringing ML models in production or creating the ML models will depend on a particular position/company. My experience is that job titles in machine learning are pretty fluid. What I was trying to do was to contrast the role of an engineer with the role of a scientist. I hope I managed to get the general idea across! :)
I agree with your last statement about ML Engineering. She implied MLE was just "coming up with algorithms" with no "building stuff". That is incorrect and worryingly moreso in a video that may be used to convince someone to choose between SWE vs MLE. Sure there will be jobs that include a Data Scientist's job in MLE but at the end of the day, it is STILL engineering. You will be building things regardless.
i'm looking to switch career paths and this video's been the most helpful yet, thanks
machine learning engineers have to be good at math, abstract thinkers, and love theories? Have you actually worked as one? BEcause this doesn't resonate with me. In all the machine learning work I've done, I don't think I've ever had to be an "abstract thinker", what does that even mean? ML work is much more about having a intuitive bone deep understanding of how ML works and how a machine "thinks" to the point of being able to know how to approach a ML problem correctly cause there is many tools in the tool belt and we have an ocean of possibilites. Thats it, that's the real differentiator between a good ML engineer and a crap one. The math involved is not that bad at all, and you only use "math" in ML to make sure you understand what is happening behind the scenes and if that's what you expected, you're not actually doing math. Loves theories? Ok, Ill give you that one, you gotta be able to sit through and understand a good amount of nitty gritty and data philosophies.
Thank you! This video was really helpful.
Especially the part about the predispositions which clarified more about the fundeamental differences between the two.
Glad it was helpful! :) I do think that our predispositions should be the key factor in picking our career paths!
Loved this video, thank you for clearing doubts, I love maths & now I could freely go for Machine Learning 😌😌
Fantastic! I'm glad to hear that the video helped to clarify that! :)
i've always heard that machine learning is about getting data-science models into production -- so like train/test splits, what are easy computational features, how do they affect your model run time, what are the run-time constraints of your model? what's your model availability constraints?
i'm not sure, how i feel about everything but i'm entering a hybrid data-science/software-engineer role at my company, and obviously it's full stack or full like our application stack so dev-ops, run time analysis, deployment, creating unit-tests, helping transform a deterministic algorithm into a probabilistic one
i think at a certain point role names are meant to try to "describe you" less than application to a certain role, just always keep learning and pushing yourself. try to focus some attention on what you find interesting, and you'll end up excelling in that area and keep getting more challenges towards that path
I believe that the first comparison is not fair as software engineering is way more established than machine learning so using percentages seems unfair/inaccurate. Other than that a good video as always
Waiting to hear your opinion on ML for CFD
You're definitely right, software engineering is more prevalent, also in smaller cities where earnings are lower and drive figures down. Averages are not the most useful statistics in general!
I'll be having a look at CFD over the weekend, chat to you soon! :)
Thank you so much for this video! I have come to realize I'm cut out for software engineering and possibly would never want to do Machine learning.
I'm glad that the video helped you realise that! :)
clear points, good delivery, nicely done. Thank you.
Thanks! :)
Please I am in high school can still go on with my dreams of studying machine learning in the future and sure of having a job
What beautiful data professionals we have today! :D I would like to thank you. You was consecutive, reasonable and probably saved plenty of my time!
Haha my pleasure! ;)
These salaries look like junior salaries especially in UK! Anything below 50k is what I would expect student or junior dev to earn in their first year!
True, especially now, and especially in London!
great content as always Karolina
you nailed it!
in my first two years at college, I knew software engineering is not what I want
I started studying Machine Learning and Data Science all by myself
last year I started my own business after only one year of studying!
now I am studying public relations and media because I think if I have both with Data Science I can get even more out of it!
one day software engineers will be replaced by AI ;)
That's amazing - to set up a business in a completely new field so fast is really admirable! How do you intend to combine public relations and data science knowledge?
@@karolinasowinska well, after what Cambridge Analytica did, and I studied about what they did, I was thinking why not doing quite the opposite! instead of changing public's behaviors and controlling what they think, why not focusing on how to save and recover reputation for governments, companies, and even celebrities. so as we know public relations is mostly about saving and recovering reputation and showing the face that we truly are and want to show, and with the help of data science and machine learning, we can collect the data we need, make NLP tools, analyze and visualize everything easier, and also combine them with computer vision. Netflix personalizes thumbnails, in our case, we can generate or at least choose templates and layout designs. it is not only the text that we can collect, it is also images, with flag classification for governments and parties, classifying sarcastic images etc...
@@keyo3945 That sounds really amazing! I'm always of the opinion that the best things happen when you combine knowledge from two supposedly unrelated areas to come up with unique insights/solutions. I see a lot of potential in what you say - good luck with that! :)
@@karolinasowinska thanks a lot 🤗, that is why I love data science, math+programming+whatever you want(music, agriculture, healthcare, media, economy, public relations...)
this is my favorite channel now so keep up the great work!(I am always here, and try to be the first to watch your new videos)
@@keyo3945 That is so incredibly nice of you, I really, really appreciate your support! Thank you, and I'll try not to disappoint with the future content! :)
Your accent it's perfect for me to understand as Brazillian
I think for your comparsion the better term would be data science, not machine learning. In my view data science is moving away from machine learning and focusing more on stats and advanced analytics. Machine learning is more and more done by machine learning engineers, because you always need software engineering to productionalize models and AutoML is getting traction.
I think you might be right. I equated the two, but the term "data science" would perhaps be clearer!
A curve can be never-ending : parabolas, hyperbolas, etc
Of course mathematically it can be! When I said that, I wasnt sure if you can have a "never ending learning curve" in English, because usually it is a "steep learning curve", and so my brain hesitated for a second!
Karolina Sowinska A curve where it might slow down eventually.. But hopefully we keep learning always!
Karolina Sowinska you are right . Steep learning curve is the right collocation.
This was very informative. I'd suggest everybody not to consider the average salary so important. Salaries don't follow a normal distribution and, therefore, the average is not very useful. For someone trying to compare salaries It's much more interesting to see entry jobs pay checks on your region and check how many job opportunities there are. In where I live, software development usually pays more than machine learning engineering but once you become a senior this trend reverses.
True, averages are typically not the best measure of central tendency! :)
i loved your video, it was very informative and gave me an idea as to what direction i would like my computer science degree to go
Hey Karolina Good video, can I ask a few questions please? well more about asking for advice, I really love math , I do like algebra and probably some statistics, planning to go to machine learning but I don't like coming up with the most efficient algorithm all the time. I like software engineering too and I love building stuff but it feels like what I learned from data structures and algorithm is kinda feel wasted? again idk why I am like this but i don't like toying around with data not to the point it's too complicated, so what should I pick if that's the case?
Although I have been doing SE for a long time, I have always to switch into ML (that I already know a lot about). This video just gave me the push.
You can do it! :)
Thank you Karolina ! this is very interesting and clear some of my doubts.
The problem is that I like both of them equally. I am creative, I am good at math, I love theories, and I am in love of coding stuff. I know Python.
And I have a choice : Go for Python ML or Go for Java SWE. And I am totally confused. This video helped me, but I still can’t choose...
In such case, you just can't go work with either. Pick one, and you'll be happy ;)
this is exactly happening to me, I need to get deeper into both to decide
I would like to clarify some of your facts. That's not how statistician looks on the data. That's data misinterpretation.
When you provide percentage as value, you need to know from what percentage it is, else you can't really compare it.
The base from the indeed report is that there was a rise from ~52 millions of ML job postings to 179 millions (344%). That means *127 mil new ML job postings*. In the Software engineering it was from 401 millions of job postings to 828 millions (206%). That means *400 mil. new job postings* in SE. Seems like scam to me, a lot more job SE job postings.
You forgot to mention the new machine learning technologies like AutoML which will probably replace a lot of not specialised (prob non phd) ML engineers as well as SE engineers. The simplification of using the ml models is real, look at GPT-3. Developers will probably be using it even though they don't have any big knowledge in math. Transfer learning (fine tuning) of existing models will also be simplified in the future.
Other than that, some nice points were shown
Fair point - I should have provided the base values, instead of just focusing on the trends - my bad! I'm aware of GPT-3's implications and even discussed them in my latest video! :) And you can expect a video on AutoML as well, I've been meaning to do it for a while. Thanks for your feedback, that's really much appreciated, it helps me to be more relevant and accurate!
The inference of salary comparison is incorrect. ML roles typically require a masters and PhD degree, whereas you'll find a lot of undergraduates in Software Engineering.
This is a very educational video, delivered in an entertaining matter. Keep up the awesome content. In my opinion, I wouldn’t be able to do both. It makes why someone would have to pick one or the other to achieve higher success.
I wholeheartedly agree that it's better to play to your strengths in life!
Karolina The Presentation Is Very Good , The Execution is Quiet Smooth and The Content you bring is easily Understood and Helpful . Thankyou ✨
I'm glad you enjoyed my video! Thanks! :)
I actually think the responsibilities of a Software Eng. and ML Eng. are not so different. Most of the time, I see ML engineers simply handling how the core machine learning algorithms are "productionized" (deployed, integrated with the application layer, etc.) so there's not necessarily a massive need to understand all the statistics/linear algebra behind the model but just how to optimize it for use with cloud/edge/hardware/iot applications. I think Machine Learning Scientists/Data Scientists are normally the ones who hold PhDs or at least masters, which are prototyping and even creating new models from scratch. Also Software Eng. deals with plenty of ambiguity as well, depending on your domain. A web dev, maybe not so much... but if you're a software engineer working with mathematical optimization or higher order algorithms, or algorithms that are the NP-hard+ space where you need to have a solver running and ingesting a massive amount of data (think large LP/QP chained model) then things begin to get nearly or just as ambiguous as the ML modeling area.
Using the definition that you're referring to, in my video I'm actually making a distinction between Machine Learning Scientists vs Software Engineers :)
Karolina Sowinska makes sense, in that case I’m actually surprised that the salary difference reported on glassdoor is not too different 😂.
Very helpful video karolina, love from India
I finished my bachelor as a Software Engineer and worked for 2 years in web and now I am doing a master's in Ai and I was wondering if those two can be combined. I totally agree that maths makes me struggle but as HeliTom said if you put effort you can make it. So what's your opinion about my question do you think someone like me with experience in business web apps and a degree in AI-Data Science could combine these two for innovative solutions
You can always combine your experience and skills in such way that something great will come out of it, so definitely - go for what you want!
Thanks you, it was really helpfull for my choices. The king of thing that get me going in the morning is software engineer.
Awesome, it's great that you understand yourself!
Could you make a video between data scientist and data engineer?
Noted, lands on my list! ;) Thanks
I'm really confused... I don't know what to do! At the moment, I'm studying at a SE University because I've always loved coding and I've always had a crush on the world of technology since the age of 10. However, when I discovered something about ML I was so fascinated... the fact that I could teach to something that doesn't have its own mind to actually start thinking and doing stuff was making me crazy and I thought it was the right choice for me! Here's the point : I hate theory. I'm a very practical person, I love coding and I love coding stuff. I'm the kind of person that wants to finish asap a manual on a programming language because I want to put my hands on the code. I love the aspect of robotics and AI. but math and theory is not for me. Is there a way to go out of this?! Hahahah
Machine Learning Engineering is the answer! :D
I've just started 4 years of machine learning. Im fine. i promise.
hahah I sure hope so!
Very interesting video. And the way you use to explain is clear. I was looking for a channel like this one so long. And finally. I subscribe
Thanks! :)
This might be the most useful/informative 12:49 i have ever spent. Thank you so much for this, you've shined a tremendous light on the path forward for me.
Also i enjoy how methodically put-together your videos are. They have an appealing logical flow :0)
Thank you again.
Wow, this is really motivating! Thank you so much! I'll keep on creating! :)
It seems that there are many machine leaning engineers that works in many different fields(computer vision, NLP, recommender system, ads ranking, fraud section in bank, etc.). Which fields hires more machine learning engineers?
Good question, I haven't seen any statistics on this though.
do you really think there's a big difference between ml engineer and software engineer 🤔 I was thinking software engineering is a umbrella term so ml engineering is just a branch and specific that's all
I meant that being a machine learning scientist/researcher is way different than being a software engineering. Sorry if I didn't make it very clear in the video!
She is right. They are very different and one is not an umbrella term for the other. But artificial intelligence can be considered as an umbrella term for ml engineering.
yet again amazingly informative, delivered with real charm
:) Thanks!
well im starting my ai-ml degree
i didnt know that it has complicated math
I liked the video. At the end I think you are contradicting yourself a little bit. If you build lots of things as a SE you will have to drop some optimality. It happens as well when running ML models. Another thing, I don't think that the practicality of SE is smth inherent from the job, but rather something to admire and even to apply in other fields. At the same time, I think SE could raise their heads and try to think about the value or the impact of what they are doing. Cheers
I am graduating next year with a masters in comp sci and am making this decision. I only know the basics of machine learning, but on the other hand I'm very good at software engineering. Because of this I find it hard to compare how much I enjoy them both as I still find ML hard to understand. Does machine learning always feel like hard work, or do you ever get to a stage where it can feel simple and enjoyable?
Everything gets more enjoyable as you're getting better at it, Machine Learning is no exception! :)
hey andrew, i'm in the exact same position as you were a year ago and would love some clarity from your experience. if you don't mind helping me out a bit, please drop your instagram/snapchat/email
perfect perfect perfeeeect comparison💛💛
Thank you so much for this video
It was interesting to listen to you)
and for me:
machine learning is more interesting and more difficult
Machine learning can be more interesting to some people for sure. And the mathematical complexity definitely makes it very difficult!
@@karolinasowinska if it is difficult with math, then you should not give up!
It's worth it!
@@denismerigold486 I like this attitude! :) Well-applied hard work is the key to success! :)
I think machine learning will become a part of software engineering and that too would done by A.I itself, But if they kept on outsourcing and Sass we will loose all the jobs. Most jobs that are done by scientists (who are also artists) will flourish, Machines can do the tedious work that we have to do over and over again but If we don’t have Scientists who define and design solutions to problems for them and tell them how to function, they won’t be able to do it. A.i. Is just a fancy name for a software tool that will do things that we aren’t good at, AGI might never even be possible, due to the complications in hardware, energy consumption by such a computer, the philosophy behind it and also you can only train the neural net to process a specific type of data set, It can’t process everything.
I think you're right! It's already happening - let's even take code autocompletion suggestions!
What if you love both? I'm a seasoned software engineer who absolutely appreciates maths and I do want to get into artificial intelligence. I've been looking for cross-functional sort of role. Can anyone suggest courses, ideas of jobs around that, please?
Hey karolina, I want to ask about the automation of which tech jobs in future and which not, especially in IT sector.(( PLEASE REPLY)) and make a contentful video with facts and figures + probabilities of happening of automation or not 👍👍👍
I think this video is very helpful. Thanks!
This is a very bad video. All the numbers are manipulated to make ML look good, eg. when she said ML jobs have 300% growth but SE has 200% that's because ML has 1K jobs wheras SE has 100K - absolute numbers tell a very different story. SE is not going anywhere and I can tell you for a fact that we SEs would be very happy if we need to code less in the future. There are a lot of different things to do in SE other than writing code.
The percentage of growth was a little misleading...there are a lot more software engineering opportunities.
True, my bad!
Thanks for the video, before I thought that the machine learning is a part of software engineering.
Machine learning engineering is a type of software engineering. Machine learning on its own (or data science/deep learning) is more of a research job :)
Growth alone as a metric is not sufficient. There are more software engineering jobs and using growth alone disguises that there were actually much more software engineering jobs opening than machine learning.
True, I should have included absolute numbers too. My bad!
Wow😲 first thank you for this video, i have stud computer science for three years and i need now to choose wether software engineer or ai engineer, the problem i face that i love both math and creative building and i decided also to get master degree after the college but i still confused with my choice, what do you think? , And thank you again for this video ♥️♥️
Very Needful Content. Thank You.
I fell in love with your videos, i was watching all of them.
With all these *SaaS* companies there is a small but considerable amount of decrease in Software Developer job role what do you think about it let me know?
I don't know if there's a decrease, but I would say that the nature of the job is changing and there is less coding and more gluing things together
This video is like boom!!! And what you are doing is agreat job 🙏:-)..and Yes please make more videos on career in machine learning,ai,dl and please help us know which are the highest paying companies in ML,Ai,dl field.
Awesome, I'm glad that the video was useful! I'll definitely make more :) Thanks!
thankyou very much mam !!! your video helped me a lot for getting clarification about both fields......................
Wonderful video. Love all your content! Subscribed. :)
Can you do an update on this video post covid? Or do you think any of your opinions have changed. Genuinely curious.
My opinions haven't changed much to be honest! But I'll need to do more research
You inspire me to pursue my career. please more video on the software engineering
I'm so, so happy to hear that! More vidoes are coming :)
You are a true ANGEL… thank you for choosing to do TH-cam videos.
Great vid, thanks! 👌✌️
Thankyou so much for making this video. Can I take what you said about SE and apply it to App Development as well for the most part? If not, I would highly appreciate if you make a video analysing App Development the way you did these two? Particularly interested in the "Your Predispositions" part which I feel like is most important.
You did a great job simplifying it concisely. Most people are too afraid of being reductionist when they simplify things that way.
I think you can definitely apply what I said about SE to App Development :) And thanks!
@@karolinasowinska Okay that's great! Any advice for someone who's just about to start (again) his Undergrad? Like is there anything you would do differently if you could go back?
why ML uses Python instead of Java?
Thanks🌹🌹🌹
Super useful video. I think software engineering sounds more up my street. Though, I do find AI interesting so who knows what i may choose down the line. Currently, doing a python coding bootcamp to get into the software engineering field as an umbrella (studied genetics so maybe that will be applicable for machine learning...)
Glad it was helpful!
Thank you so much 💖
Excellent comparison! Happy subscriber. 😊
I'm glad you enjoyed it, thanks for being here! ;)
I have heart and read on social media that Data engineer earn even more than data scientist or machine learning engineer, is that true?
It really depends on many factors such as location and industry!
Mam, please make a video from strach to advance how to become a data scientist?
Thanks for the suggestion, I have a few similar videos on my channel, do check them out! :)
Great video! That clarifies a lot of things. I just have a couple questions - does it make much difference if you have a BSc followed by an MSc as opposed to a 4 year integrated masters (MSci)? And do you think a (separate) masters in ML is a good idea?
I just got my first machine learning job... I'm an electrical engineer, but it just kind of worked out
Do you really need a masters to get a job as a machine learning engineer? I mean like now there are tonnes of online courses out there which you can do parallel with you bachelors...?
Hey, no, Masters is not a requirement! But quite often this is what employers explicitly ask for in job openings.
Ohk, so it gives you an edge over the other applicants (that said if they don't have a masters)?
@@arjunb4241 Yes, you can put it this way!
@@karolinasowinska ok 👍
your videos are so helpful and pleasant. thank you!
I'm glad you think so! Thank you! ;)
Thanks Karolina
Very Interesting Information.
I'm glad you think so, thanks! ;)
Thanks for the video. Can you please tell which masters degree is good for machine learning.
Any "Machine Learning" or "Data Science" Masters degree from a good university :)
What if I cant answer these questions at the end? I have to decide in a year
Hmm... maybe come back to them in a few months :)
Who is paid more software architect or machine learning engineer?
very nice explanation.
thank u so much
Karolina, when you say SQL, is better MySQL or PostgreSQL? Any thoughts about it? They will be well received
A growing segment is data engineering. The need someone to ingest the data into the cloud before they can used ML to analyze it.