🚀 There is so much more to explore in ML. Feel free to grab my FREE cheat sheet of different ML domains and open challenges: borismeinardus.substack.com/p/a-list-of-different-ml-domains
Thanks for making such videos on becoming a ml researcher. Your content has made me more passionate for ML research , particularly in deciphering and understanding research papers
Great message. I think videos like these really help the community by giving important context and a reality check on a research career. I'd also love to know your thoughts or approach to navigating career choices, e.g., staying in academia vs working in an industry lab vs pure industry (applied).
Really glad you enjoyed it! In general, I think working in an industry lab is better than a university lab. You get more money, arguably work with smarter people, probably have access to more compute, and pretty much do the same work. How to pick what suits you best is of course not that simple to determine via a comment haha I will hopefully soon create a video exploring the different ML roles. It's a huge mess haha every company is using whatever title they think makes sense. It's easy to get confused.
Yeah, trying to find a good balance to help people who want to be an engineer as well as researchers. That said, I do believe that most things I say apply well to both paths ☺️
As someone who decided against a PhD in favour of becoming a better software engineer after my masters, i'd be interested in your take on the Ph.D. vs. non-Ph.D. debate. Personally, I felt that writing production level code was more of a hurdle to me than the experiment driven development of ML models. So I thought I would learn more in the same time becoming an swe as opposed to researcher. Currently, I work as an SWE in MLOps and still think that (at least for industry) its more valuable to be able to train and deploy mediocre models than to come up awesome ones that you can't serve or even train because of skill gaps. Yet, who wants to build mediocre models, so the idea of still pursuing a PhD comes up every now and then. Time will tell :)
Hello brother, I will be starting my masters in data science in upcoming fall. I have already gained some skills through my projects but now I am deciding to delve completely into this field. Your videos have been and will be really helpful throughout this journey. thank you for indirectly encouraging me.
The past few days, I've been considering whether I should pursue a research career in ML. Your video gives me a new perspective on what to expect. Thank you a lot :)
I'm really happy that the video could help a bit with your decision making! I would still recommend to do some further research 😊 Best of luck and have fun!
Best of luck my friend, the path sometimes is hard, but without the hardness we cant become great, again best of luck, and thanks for such useful content!
How have you found ML within academia? With code etc not being as neatly needed/presented for papers have you found a difference between academia and industry?
There are a lot of very poorly written repositories, but once you find one that is well written you can build on top of that very well. These good repositories are then again often from industry research divisions haha. E.g. I use Salesforce 'lavis' repository and can highly recommend it :)
So as someone who's getting a master's in AI... would you say that it's just simply easier for me to go for a software dev role post graduation, instead of trying to go for an applied scientist/ml engineer role?
How do you get research experience if you don’t plan to pursue a PHD? Also what if you like to work on a wide set of problems rather than just one problem for a long time ?
The beginning of pretty much always the same. Perhaps have a look at my how I would learn ML video. From there on you start to read more and more papers and start to think of how to write one yourself in a topic you find interesting :)
Can u Please make a video on: "How to overcome PHD students being a Bachelor student at Big Tech FAANG companies".... btw Love to Watch your video.... you are doing great bro....🙂👍
How to b a ML Researcher without a PHD in a Big tech? Is it even possible? How should I b developing my resume so that I can secure an ml Researcher position in MAANG? What should b my dos and don't? Can u plz make a video about ur opinion on this matter, whenever u have time. Have a nice day, Sir.
The short answer is: it is possible but it's hard and arguably messy. Either way, in general, you will have to spend at least the same amount of time on the right career path in industry to get a research role without a PHD, as if you would have spent on the straight forward path in a PhD. But I will think of a way how I can create a good video covering these questions :)
Hey Boris, thanks for the video. I had a question regarding lateral transition to a researcher role. What do you think an experienced ML engineer(say 5 YOE) with a few average research paper and IPs in the field should do to be considered for a position of research scientist at these up and coming AI startups and established research labs in big tech companies?
How does a PhD in non machine learning degree sound like? I'm doing my PhD in Chemical engineering and it is a mix of experimental and modelling work. The modelling is mostly machine learning and CNN. Would my PhD be as good as someone who has a PhD in machine learning only?
haha thank you ☺️ doing my best to find the best balance between my job and posting videos. But I post more often is in my weekly newsletter :) perhaps that might be interesting to you ☺️ Thank you for your support 💛
Ah man, I knew this might come haha arxiv.org/abs/2311.16829 We submitted to like a B-Tier conference. Please don't judge last years Boris too harshly haha
@@borismeinardus Ahh don't act like that...I know you want it haha, nawh don't worry I would never judge because of such a research paper, since even not accepted it's a huge commitment. Thanks for sharing! im looking forward to read it and gain some new knowledge
Boris bro! I’m working on my master thesis and I’m desperated. The topic: optimization of process parameter through machine learning Context: it’s a production line, with sensors. I will get the data and have to find the optimal parameter for the prudction line. Such as temperature, vacuum or the membrane. Which model would you use?
I'm also not quite sure if ML is the right thing to use here. You would somehow need labels that represent good performance, which then means you don't need to solve the problem anymore. I would also say try out some sort of grid search and perhaps try to build some sort of simulation that predicts the performance given the set of parameters. For this you might use very simple ML to learn a model that maps the input parameters to a certain output metric you have the labels to.
He said each time a carbon product is finished they do a 3d Scan. In there they can see the results. Which one were good and which one are bad and have to be recycled or something like that. Maybe I can choose this as the Y variable somehow with the given input variables which I can edit?
should work as a regression based model. You just need to figure out how you can best select and represent the input features. But I think you won't need a really complex Neural Network for this. This project sounds to me like a more Data Sciencey project, where it is more about understanding the data rather than developing a complex NN (perhaps (probably) don't even need a NN here).
Hopefully, as I wanna successfully finish this off. I decided this topic in order to get into the programming world. So basically linear regression model, and checking out the correlation, to see which input data has an impact. Also do you know how I can basically take the 3d scans of each product with those parameters as the Y variable? The original idea was actually a close loop. So the system can regulate the parameter by itself.
