ไม่สามารถเล่นวิดีโอนี้
ขออภัยในความไม่สะดวก
What Professional Machine Learning Engineers ACTUALLY Do
ฝัง
- เผยแพร่เมื่อ 1 ส.ค. 2024
- Most people will only show you the highlights of their job, today we will look at what professional Machine Learning Engineers actually do in their life.
We will not only look at the many positive things Machine Learning Engineers experience but also the not-so-positive side, and be honest. And I will share many things you don't know about the job and what differences the many data roles may make to this.
If you enjoyed this video, I would be excited to connect on Twitter or LinkedIn.
Twitter: / datawithsandro
LinkedIn: / sandro-luck-b9293a181
Medium: / datawithsandro
Intro: 0:00
What is it actually like to be an ML Engineer: 0:30
Not only the positive: 1:15
What your average day will be like: 2:14
Differences between the roles: 4:30
What you don't know about the Job yet: 5:51
00:00 Most videos about the life of a machine learning engineer show only the highlights, but the reality is that they spend most of their time behind a desk, working on a computer.
00:59 A typical day for a machine learning engineer involves a morning stand-up meeting with the team, discussing problems and plans for the day. There may also be additional discussions and collaboration with team members on infrastructure and project-related issues.
03:27 The day is usually divided between meetings and coding, with meetings involving discussions with stakeholders about model performance, technical discussions about infrastructure, and planning for future projects.
04:53 The job titles in the field of data can vary greatly, and it's important to understand the different roles and levels of variation within a company before joining.
05:23 The distribution of tasks in a machine learning engineer's day can vary, with a mix of machine learning project work, infrastructure work, support, and meetings. These proportions can fluctuate depending on the stage of a project or other factors.
06:22 Machine learning engineers often work on the same type of ML problem for a long time, focusing on a specific domain or use case. They also work on someone else's code base and tend to settle on specific tools and technologies.
07:51 Documentation is an important part of the job, including writing documentation in Confluence, writing emails and messages, and documenting progress and decisions in JIRA.
08:51 Real-world data is often messy and noisy, requiring a lot of data cleaning and dealing with unexpected issues. Finding clear correlations and solutions can be challenging, and many projects may not have the desired outcome.
10:19 Despite the challenges, being a machine learning engineer is still considered one of the best jobs in the world, but it's important to acknowledge the imperfections and not pretend that everything is perfect.
Thanks for saving my time👍👍
Extremely helpful, thank you 😊😊
loving the 'slightly more honest' quip Sandro!
I was looking for a video about a realistic day to day of AI/ML engineers, thank you!
You're very welcome!
A very true video... Really appreciate it 👍
these video series are great content. I am coming from an academical background and lucky to had internship on ML/AI and now with your videos I had a chance to get an insight look. Thank you.
Lovely to hear that you will join the professional world, and very happy it was helpful🙏
Came to this video right after wasting 3min on another video just like you described in the beginning. Thank you!
Hey Sandro, I am looking for a piece of advice. I am a mid level automation QAE in a large company. I want to believe I am pretty technical and in my role I have a lot to do with coding. I’ve been working on transitioning into SDET role, but with AI becoming a the biggest deal in AI I am considering taking a ML bootcamp and trying to land a role as ML SDE. I don’t have a computer science degree and learned everything myself and by working. Do you think it’s realistic to get an ML Engineers role with my background after taking a comprehensive bootcamp?
What is the value of even trying to solve these real-life problems if, most of the time, there is no solution, no correlation? If my work is not being used, will I risk getting fired?
Good question, maybe the video is a bit too harsh from that perspective.
While very often the correlations that everyone hopes are very strong are a lot weaker than expected (due to various forms of noise, or just generally people being quite optimistic about what they do next). The correlations everyone trivially expects to exist have usually already been exploitet, which led me to the statements in this video, it's not that everything is random, it is just that lot more noise exists in the real world then we would expect. For the value perspective: it is often a lot of value gained, and process optimization or next steps that can be inferred from your work, also the outcome current data may not be enough is a good outcome, because you can optimize the process for your next evaluation in 6months, and killing bad projects early or rephrasing them is valuable in itself. Additionally when dealing with multiple millions ( as most business hiring data scientists etc. do) a few .% are already worth a lot😉
Please make video on all stages in ml project and tools u use.
Sounds great thanks for the technologies input!
So tough to break into ML tried to land a job for over a year now just given up on it 😔
Do you mentor people? Like one on one mentoring to help those wanting to get into the field?
Well, i got invited after entrance test to enroll to University of Oulu for CSE program undergrad + grad and planning on spesializing in AI & ML. Seems like theres many videos to watch now :)
Great to hear!
Hey can u make a real time project with image classification one? Like dockerzing, using gcp or was for deployment everything?
Sounds actually quite interesting, however i feel this will be more than one video and more like a series. I will think about that direction, thanks for the input
@@datawithsandro2919 hey if we could connect in linkedin.. I'm a fresher have some experience in few projects would love to work with you for a project that may put an impact to the society.
I have sent you connection request.
Hello im currently on my last year of bachelor in cs .i m goodan pythonand its few ml libraries should i focus on learning math and ow to build model or sql, data cleaning andhow to use model please help 1:58
Yes. You're welcome.
hey getting offer of engineering AI/DS AND AI/ML what should i choose ??
Have a look at the company the position titles can be somewhat misleading, is the only tip I can give the rest really depends on what you like the most
@@datawithsandro2919 Thanks for your help 😃!
Can you make a video on MLOps please
I was trying to put something together once, and realized it takes several hours. Once I find a good way to break it down into manageable chunks I will for sure 😊
Hi Maestro, hope you are doing well. I just wanted to ask, if I do masters in computer science...can I have a prospect of becoming ML engineer in the future?
P.S. I believe that without becoming a good software engineer, I can't become a good ML engineer. So asked.
Yes, you will. It'll actually give you an advantage
Isn't AI engineer and ML engineer seperat major or ther is like mix of them
Depends heavily on the area/ company you are in, Ai Engineer might be used more in consulting or startups ( I don't think it is used in many companies as a job title). In the end it is developing topic
I'm getting my BS in data science currently and will be moving on to get my masters in machine learning. I'm not doing it for all the "fun things" in fact I would dread having to do those things that people consider fun. I'm pursuing this path because I want to sit behind a desk for 6 hours a day (preferably remote work from home) because I've had jobs where I had to wake up at 3 in the morning drag myself to work just to work in the rain at 15 degrees Fahrenheit.
How hard has it been for you?
As MLE do you do model building ?
I would say yes i do, but it depends a bit on what 'model building' means. Do i design fully new Ml architectures from scratch -> no. Do i build ml systems that use multiple state of the art algorithms and combine/train and put them into production -> yes. For the rest in between i would say sometimes
@@datawithsandro2919 Very honest answer. designING fully new Ml architectures from scratch daunting but costly in terms of not only finances but manpower also
It’s amazing how this video: What professional software engineers ACTUALLY do, th-cam.com/video/Q0A35ZfgwHA/w-d-xo.htmlsi=1GUqy1dK-dzVuHX0 talks about the exact same thing. Exact same words, exact same points …
I’m not sure who copied who, but how about being just a tiny bit more creative with the copied content ? 🤦🏾♀️
Oh scratch that, I now know who copied from whom 🤦🏾♀️🤦🏾♀️