One of the best “A day in the life” videos I've seen and actually provides insight into the work! Not like many others, just vlogging their day and joining meetings without sound all the time. Love it!
You can buy a used laptop to browse the internet and do stuff while your model trains. Or buy one of those mini-PCs with Oculink egpu support to use as a training system.
Great video! As a Data Scientist working remotely for a gaming company, I felt nostalgic for the days when I used to gather with my colleagues in a small room to work together like you guys! Keep going!
Man, from watching you doing coursera courses now building at scale, time goes by. Same goes for me, from being a school student to working in an AI startup. Ur chanel has been an inspiration for the idea of learning
Now THIS is what you call doing what you love. It felt very real and authentic. Nothing was forced yet it was pretty organized. I actually had no idea what went through Machine Learning and it turns out it is A LOT but you make it look easy lol. Subbed.
I worked at a similar small startup recently. Just me (ML), the sales guy (founder) and the infrastructure guy. Unfortunately, we didn’t get the funding needed to keep it running so we had to close shop. I really loved it while it lasted. Why? Everyone on the team could code, in fact I was the least experienced guy (the others each had more than a decade of experience in Java). I was working primarily in Python. Whenever I got stuck, the sales guy will come in to help. I felt blessed working with people who knew what’s up. I didn’t need to dumb down things for them to grasp.
If I will be able to work with this man in such a balanced setup, trust me I will work all the time. I am learning deep learning while studying physics from 3rd world country. it's hard to understand by how much we go through living in a third country. I wish I will make it to go to a developed country!
good work mate.I had this idea of blending health and AI for a job, but I didn't go all in. Now I'm shooting for a corporate role as an MLE, and props to that bold dude who took the plunge into starting his own business!
I admitted that I "paused for book spotting" at 0:55, and nice to see many captions/emoji around. When you use jargons, there are explanation on the screen. Yes, I will pause and read it.
Love the video, proper insight into what your doing unlike the typical tech day in the life vids. Fills me with inspiration to emulate what you guys are doing in the future.
i feel like a lot of youtubers make it seem that mle is building models from scratch when in reality the models are already built. Most of the work simply seems like it goes into data collection
I think the most important metric to improve is cycle time for experiments. Two questions: 1. Why not use ChatGPT / Claude / etc. to label pictures and review the labels 2. Why not use the Cloud to train the models Would reduce the training time significantly and you get feedback a lot faster. Drawback: Costs
Love love love this video. Thank you for uploading. Esp love that you're training your models locally and also the data-first approach. Keep up the good work!
as a former music producer/musician I was looking for inspiration for coding cause i'm currently learning it. this video is awesome in this aspect. it feels like if you maintain the vibe, the curiosity and all the progress is a really exciting way of living
Ahhhh love the vibe! Also, finally someone doing ML Engineering version. Keep it up more mate. I'm also an ML Engineer for a while and really love your vlog.
This is a true day in the life blog! thanks for sharing and mentioning the technical details and roadblocks in the video. super impressed with the progress of the app and will be following along with the journey
Familiarity with the dataset you are working with is crucial for an ML engineer so kudos for getting your hands dirty and labeling. Also, if I had to look at food images all day my body weight will explode like my gradients. Thank you for sharing!
Thank you for showing me this. It really fascinating, it looks really authentic, I am so excited to learn machine learning. Thank you so much guys, God blessed you.
Hey I just wanted to say this was a great video. I want to become a machine learning engineer and I feel like this one of the few day in a life videos that gave me an idea what to expect
@@mrdbourke hey! revisiting this video today for more insights...kinda loved the complete explanation of the process, looking forward to more similar vlogs
I just feel happy watching this video. I still remember how you started out learning ML and DL in 2019 lol. You were the inspiration for me as I also started out around the same time lol. Fast forward 5 years and here you are with your own start up! Truly happy to see such development!
You have truly inspired me, and I wish you the best of luck in your endeavors. I watched the video several times, and I enjoyed the small work environment and the interaction between you. I also liked how you manage your time effectively, eating, resting, exercising, and reading. This discipline will help your company achieve its goals because you have the necessary qualifications and capabilities.
Hey, please update the Machine Learning Roadmap, you said you'd post a new one in 2024, I'm waiting for it. Also, whilst learning, I'm sure you encountered moments where your code didn't work for some reason, how did you resolve the issues, and sometimes you don't even know which part is the problematic one.
Yes you’re right! It is time for a new roadmap! Also, I get code wrong all the time. Like even in this video there was an error with dimensionality and my model training failed. Programming is mostly writing bad code and then fixing it over time to make it better and better. When I get something wrong, I try to reduce it to the smallest point of reproducibility to fix it. E.g. keep narrowing it down line by line until you find the bug.
This video made me jealous. I am learning machine learning for like 3 years, hustle a bit, did some online unpaid projects, eventually got into a job, but in my job we only uses pretrained models api for various work. man i need to go out and train models. I am missing all the struggle and fun.
God damn it! Take a sub! This is the best day in the life of video with actual substance and not just fluffing about. Well done. Great to see another Aussie startup!!
It could be, I haven’t experimented with it myself. Maybe could rent out a large cluster for an hour or so for training runs. But I like the convenience of having a GPU station on hand.
@@mrdbourke electricity bill vs cloud bill - how important it is to run these experiments for pmf and growth, just careful with the vision modals outrunning the show, hardest job here seems to be the ios guy
Daniel, thanks for the video! You could make a video about what it's like to create a startup, how to get customers and everything related to a startup.
2 questions 1. why are you using weights and biases instead of ml flow or tensorboard? is it because monitoring is better on weights and biases?? 2. why do you use GCP? i see you are training model locally, are you using gcp to store model artifacts?
This is so relatable. Experiencing the same routine. As far as fusing of vision and language model is concerned, you should probably use a VLM eg Florence 2 or Paligemma
I took a team through ML modeling this summer and I liked how you use the food items that go into the labeling process then processes the imagesets through linux. Have you gotten around the validation step that confirms the labeled images specifically the data in them can transfer to produce a trained set. that data can convert to a new container for specializing the data like here. You guys have the Nutrify application as a output source what about a dataset analysis that confirms the age of fruit?
Oh man! I was randomly watching the video and enjoying the good vibe at your workplace. Then I noticed the name Daniel Bourke. You and Andrei Neagoie made an Udemy course which made me who I am today! Thank you, guys!
This video was helpful to me as an mle myself. Nice to see your models training in real time. Cool stuff. I need to brush up, but hire me in six months 😊
Hi, software engineer here from OC California but born in Toowoomba, back of Bourke not far from you. Got into ML back in the 90’s after reading issues of AI magazine. I interviewed at an office that looked remarkably like yours in OC but this was the time of Windows 3.1 and Unix systems 5. These guys had created their own analog neural network boards and had these minicomputer chassis hollowed out with them in, sans tensorflow, no graphics cards, very tin cans with a string. Plus the activation function issue was not handled at this point in time. Often wonder what became of them.
Very very nice idea, most likely already implemented, but perhaps you guys can mess around with recipes and perhaps even help with people’s cooking with your object + judging food quality/ foods interpretation with local LLMs(can be more expensive, but your solar power source can shoulder some of that!) when you guys get your object detection down. Loved the design as well. perhaps web scraping images can speed up the object detection collecting but I’m sure that leads to some cans of worms. Would love to see how this project goes and it boost off!
ML is quite the big field! I’d say one of the best places to get started is to learn Python programming and then start building some models of your own. I’ve got a blog post here about how I got started: www.mrdbourke.com/startaiml/
what i have noticed with my short use of this programs is that it doesn't weigh the pros and cons of certain things accurately to give you what i believe is the actual goodness rating
This is the best "A Day in the life" video which i have watched so far. literally loved the whole video and how u have shown what ML folks really do. Kindly make more of such videos.
Creating a solid business plan doesn't have to be complicated. With Corvus Link, you can simplify the process and turn ideas into actionable steps to bring your startup idea to life.
Very insightful, the actually helpful kind of "a day in the life of" some kind of profession. Really helped me to get to know more about how an ML engineer works.
I can think of several reasons: 1) Your own PC is just that: you can do whatever you want with it and whenever you want. 2) In case you didn't know: whatever runs on Google, stays on Google. So unless you want to share your company's trade secrets with Google, get your own PC. 3) The GPU's provided by the free version of Google Colab aren't that good and also get disconnected quite often. Buying the pro version gets you a better GPU and a limited number of computing units to use it. If you're running models 24/7 (as mentioned in the video), those units will run out very quickly so you'll have to buy more all the time. Best to just get your own setup and just deal with the electricity bill, which should already be handled by the solar panels these guys have on the house.
Reminds me of a very popular tv series, 5 years ago..with the prediction of LLM's ...btw very Aussie style with the food the exercises and the sun, man I miss the sun u get there,
You forgot the 1 hour meditation, the audio book, the cold plunge, the sauna, the 1 hour gym session, 2 hours of answering emails.
😂
@@mrdbourke lmao,ubuntu? nah mate you lame
@@mrdbourke i use arch btw
The 2 hours of answering emails struggle is real
You forgot the Master Course we can all buy for $100
This was the best "A Day in the life" video so far, very authentic and felt like being part of the team 😂
This is the video where I felt like I want to be like them! Productive. Enthusiastic.
" felt like being part of the team" 🤣
Also to note they have their startup in their home too feels more relaxing than a commercial space
i love how authentic this video was, none of that drama shit
Glad you enjoyed! We like to keep it pretty chill and simple.
One of the best “A day in the life” videos I've seen and actually provides insight into the work!
Not like many others, just vlogging their day and joining meetings without sound all the time. Love it!
Being an ML engineer with no source of income is not a joke. I am kinda new into ML and I trained my model for a day and couldnt use my PC
You can buy a used laptop to browse the internet and do stuff while your model trains. Or buy one of those mini-PCs with Oculink egpu support to use as a training system.
Is a virtual machine
not a good approach?
@@rikudouensof even the virtual machines cost money 💀 I'm cooked
@@roast-salamander oracle vm is free?
Run command in background mode
Great video! As a Data Scientist working remotely for a gaming company, I felt nostalgic for the days when I used to gather with my colleagues in a small room to work together like you guys!
Keep going!
Data science for games? Interesting.
Man, from watching you doing coursera courses now building at scale, time goes by.
Same goes for me, from being a school student to working in an AI startup. Ur chanel has been an inspiration for the idea of learning
No day in a life video matches this on. Dude actually shows his code as if he is doing a tutorial. This is just wonderful man, God bless your hustle🙏
Thanks again for creating that massive PyTorch course. That was a great help in my journey into ML.
Glad you liked it! And you're very welcome! Enjoy the epic world of ML
@@mrdbourke Hi! Where can i find said course?
@@tomhoq2003 on my RUclips channel (search “Learn PyTorch in a day”). Also see learnpytorch.io
This video made me happy, i have been lerning data science for the last 6 months and got a bit tired, this video gave me push, thanks man
Now THIS is what you call doing what you love. It felt very real and authentic. Nothing was forced yet it was pretty organized. I actually had no idea what went through Machine Learning and it turns out it is A LOT but you make it look easy lol. Subbed.
I worked at a similar small startup recently. Just me (ML), the sales guy (founder) and the infrastructure guy. Unfortunately, we didn’t get the funding needed to keep it running so we had to close shop. I really loved it while it lasted. Why? Everyone on the team could code, in fact I was the least experienced guy (the others each had more than a decade of experience in Java). I was working primarily in Python. Whenever I got stuck, the sales guy will come in to help. I felt blessed working with people who knew what’s up. I didn’t need to dumb down things for them to grasp.
Wow! That's how ML development is being carried outside of school. Now, i'm hooked!
Woohoo!!! Word of warning: ML can be very addictive :P (I'll let you decide whether that's good or bad haha)
@@mrdbourke Definitely bad :P
If I will be able to work with this man in such a balanced setup, trust me I will work all the time. I am learning deep learning while studying physics from 3rd world country. it's hard to understand by how much we go through living in a third country. I wish I will make it to go to a developed country!
living in a 3rd world country here too, bro. It sucks lol
where?
Stay home fix your own country
@@Soulseeologia get your point... but not realistic haha
@@darianmoratExactly
good work mate.I had this idea of blending health and AI for a job, but I didn't go all in. Now I'm shooting for a corporate role as an MLE, and props to that bold dude who took the plunge into starting his own business!
I admitted that I "paused for book spotting" at 0:55, and nice to see many captions/emoji around.
When you use jargons, there are explanation on the screen. Yes, I will pause and read it.
Love the video, proper insight into what your doing unlike the typical tech day in the life vids. Fills me with inspiration to emulate what you guys are doing in the future.
Sick idea. Good on you for seeing this vision and keeping at it.
This was actually a genuinely good video, which I'm delighted to have watched. Please, do keep on making these.
i wish u all the very best mate. Definately will remember u.
Wish you guy all the best success
This looks like a path to be on, not stuck in a shitty cubical job
Thank god the algorithm recommended me this, content is pure gold!
This is the most productive Day in the Life video in all aspect by far. Wish of luck to your goal as a startup. All the best
Half way into the video and learned a lot more about machine learning than I did elsewhere. I love the positivity and great vibes. Keep it up guys!
That’s a big compliment! Thank you and glad you’re enjoying
These 3 genius people in the small room is working on an app that can change the world. So amazing! I was inspired!
I’m all for these guys success…but change the world? lol gimme a break
i feel like a lot of youtubers make it seem that mle is building models from scratch when in reality the models are already built. Most of the work simply seems like it goes into data collection
I think the most important metric to improve is cycle time for experiments.
Two questions:
1. Why not use ChatGPT / Claude / etc. to label pictures and review the labels
2. Why not use the Cloud to train the models
Would reduce the training time significantly and you get feedback a lot faster.
Drawback: Costs
감사합니다.
Love love love this video. Thank you for uploading. Esp love that you're training your models locally and also the data-first approach. Keep up the good work!
best 'day in the life' video by far. Wishing you the best with everything. Good luck.
Very kind words! And thank you 🙏
as a former music producer/musician I was looking for inspiration for coding cause i'm currently learning it. this video is awesome in this aspect. it feels like if you maintain the vibe, the curiosity and all the progress is a really exciting way of living
This is really awesome! Love how grounded and straightforward you explained everything.
Cool vibes and realistic 'A day in the life'...
Keep ML-ing...
Ahhhh love the vibe! Also, finally someone doing ML Engineering version. Keep it up more mate. I'm also an ML Engineer for a while and really love your vlog.
This is a true day in the life blog!
thanks for sharing and mentioning the technical details and roadblocks in the video.
super impressed with the progress of the app and will be following along with the journey
this is the best startup video on RUclips
Impressive stuff Daniel. Watching a dedicated ML/CV developers' team on site teaches a lot, very motivating, thanks!
Familiarity with the dataset you are working with is crucial for an ML engineer so kudos for getting your hands dirty and labeling. Also, if I had to look at food images all day my body weight will explode like my gradients. Thank you for sharing!
Thank you for showing me this. It really fascinating, it looks really authentic, I am so excited to learn machine learning. Thank you so much guys, God blessed you.
Hey I just wanted to say this was a great video. I want to become a machine learning engineer and I feel like this one of the few day in a life videos that gave me an idea what to expect
You got this! Being a machine learning engineer is all about being one with the data and keeping the data flywheel going.
LOVING the enthusiasm and passion
Thank you for being authentic and no cinematic bullshit
This was the best one so far man, appreciate it!
Thank you!
Brother, whatever you're feeding your brain, keep it up! Your insights and energy are next level. Loving the content!
Thank you! The secret is whole foods, not too much!
Awesome bro. Hope this succeeds. Been wishing something like this existed!
This was really a day in the life video, really appreciate your works.
Thank you! Glad you enjoyed it :)
This video legit made me download your app and start using it.. Well done guys!
Woohoo! Have fun and let us know if you feel we could improve it, we're all ears!
I feel the enthusiasm. God speed bro! Wish you the best
The error explanation notes are genuinely amazing, might be one of the reasons I continued watching
Glad you liked them! I realised when editing how many things I said that someone who's not 100% familiar with ML might not fully understand.
@@mrdbourke hey! revisiting this video today for more insights...kinda loved the complete explanation of the process, looking forward to more similar vlogs
I just feel happy watching this video. I still remember how you started out learning ML and DL in 2019 lol. You were the inspiration for me as I also started out around the same time lol. Fast forward 5 years and here you are with your own start up! Truly happy to see such development!
Woah! Thank you for the support over the years. Hope you're well brother. Keep machine learning!
You have truly inspired me, and I wish you the best of luck in your endeavors. I watched the video several times, and I enjoyed the small work environment and the interaction between you. I also liked how you manage your time effectively, eating, resting, exercising, and reading. This discipline will help your company achieve its goals because you have the necessary qualifications and capabilities.
Wow, what a beautiful comment. Thank you so much for the kind words, we really appreciate it.
this is so fun to watch and motivating that i feel like working with you guys or build an awesome team like yours
wow! man great work with the app and machine learning
thank you for the great content
you guys doing thing, doing really good things. This is what i think a real developers
Have been following you for years.happy to see the progress.i feel so lazy
home office, i love it. my training machine only a 7 years old 2nd hand note book , gtx1050 with cuda. it just fit my need!
Hey, please update the Machine Learning Roadmap, you said you'd post a new one in 2024, I'm waiting for it. Also, whilst learning, I'm sure you encountered moments where your code didn't work for some reason, how did you resolve the issues, and sometimes you don't even know which part is the problematic one.
Yes you’re right! It is time for a new roadmap! Also, I get code wrong all the time. Like even in this video there was an error with dimensionality and my model training failed. Programming is mostly writing bad code and then fixing it over time to make it better and better. When I get something wrong, I try to reduce it to the smallest point of reproducibility to fix it. E.g. keep narrowing it down line by line until you find the bug.
Thanks, will be waiting for the updated roadmap.
This video made me jealous. I am learning machine learning for like 3 years, hustle a bit, did some online unpaid projects, eventually got into a job, but in my job we only uses pretrained models api for various work. man i need to go out and train models. I am missing all the struggle and fun.
God damn it! Take a sub! This is the best day in the life of video with actual substance and not just fluffing about. Well done. Great to see another Aussie startup!!
wonder if it'd be cheaper to do the training in AWS Cloud, and certainly cut down time from 11-12 hours, to 1-2 hours (10x).
It could be, I haven’t experimented with it myself. Maybe could rent out a large cluster for an hour or so for training runs. But I like the convenience of having a GPU station on hand.
@@mrdbourke electricity bill vs cloud bill - how important it is to run these experiments for pmf and growth, just careful with the vision modals outrunning the show, hardest job here seems to be the ios guy
Daniel, thanks for the video! You could make a video about what it's like to create a startup, how to get customers and everything related to a startup.
Super nice office, the backyard is super nice to take a pause or unwind after work with colleagues!
Even if it is for promotion, it is a very refreshing "day in the life" !! Good work ! :)
2 questions
1. why are you using weights and biases instead of ml flow or tensorboard? is it because monitoring is better on weights and biases??
2. why do you use GCP? i see you are training model locally, are you using gcp to store model artifacts?
I hope that one day I will have a hard-working and harmonious team. Good work !
Really cool video, guys! Keep pushing hard! I have a data startup here in Argentina, and most of my days are very similar to yours
Thanks for uploading a video...I have subscribed to you 5 years back
Informational, wholesome and genuine record of your day. I enjoy the video a lot.
This is so relatable. Experiencing the same routine. As far as fusing of vision and language model is concerned, you should probably use a VLM eg Florence 2 or Paligemma
I really enjoy watching your videos. Keep going, guys
I took a team through ML modeling this summer and I liked how you use the food items that go into the labeling process then processes the imagesets through linux. Have you gotten around the validation step that confirms the labeled images specifically the data in them can transfer to produce a trained set. that data can convert to a new container for specializing the data like here. You guys have the Nutrify application as a output source what about a dataset analysis that confirms the age of fruit?
love the video, very authentic "A day in the life "
Oh man! I was randomly watching the video and enjoying the good vibe at your workplace. Then I noticed the name Daniel Bourke. You and Andrei Neagoie made an Udemy course which made me who I am today! Thank you, guys!
So stoked to hear legend! Glad you enjoyed the course :)
What a wonderful content! Keep going, surely its going to be a banger!
Thank you! We really appreciate it.
Yes, I love this. Thanks for sharing. I'm working on setting up an ML lab in my home office and this is sooooo relatable.
This video was helpful to me as an mle myself. Nice to see your models training in real time. Cool stuff. I need to brush up, but hire me in six months 😊
Absolutely love this day in the life! I aspire to start something in the future, and this was super motivating!
You got this!
As a future ml engineer I find this video inspiring 😌
Love the exercise snacks. I'll put them into my routine! Thanks for sharing.
Hi, software engineer here from OC California but born in Toowoomba, back of Bourke not far from you. Got into ML back in the 90’s after reading issues of AI magazine. I interviewed at an office that looked remarkably like yours in OC but this was the time of Windows 3.1 and Unix systems 5. These guys had created their own analog neural network boards and had these minicomputer chassis hollowed out with them in, sans tensorflow, no graphics cards, very tin cans with a string. Plus the activation function issue was not handled at this point in time. Often wonder what became of them.
probably the best "a day in the life video" I've seen
I would be hungry all the time just by looking at all those pics. 😂😂😂😂
Very very nice idea, most likely already implemented, but perhaps you guys can mess around with recipes and perhaps even help with people’s cooking with your object + judging food quality/ foods interpretation with local LLMs(can be more expensive, but your solar power source can shoulder some of that!) when you guys get your object detection down.
Loved the design as well.
perhaps web scraping images can speed up the object detection collecting but I’m sure that leads to some cans of worms.
Would love to see how this project goes and it boost off!
You look so happy I love it. I wanna get into ML don’t know where to start tho
ML is quite the big field! I’d say one of the best places to get started is to learn Python programming and then start building some models of your own. I’ve got a blog post here about how I got started: www.mrdbourke.com/startaiml/
what i have noticed with my short use of this programs is that it doesn't weigh the pros and cons of certain things accurately to give you what i believe is the actual goodness rating
Hey there, thank you for the feedback, do you mind expanding a bit on what you mean? Could you give an example workflow?
This is the best "A Day in the life" video which i have watched so far.
literally loved the whole video and how u have shown what ML folks really do.
Kindly make more of such videos.
Glad you enjoyed! Thank you for the kind words
This is a really healthy a day in a life. Thanks for making such great content
Thank you! Glad you enjoyed :)
For a image classification model won’t recall and precision be better metrics for gauging the performance of the model
Awsome 👍 video Daniel good knowing how Ml engineers work in day to day and also workflow optimization!
sound is low.
Excellent annotation.
Also great split screen effect.
An incredible video brother. Felt like part the team ❤.
That was a neat little trick to leave the PSU outside the case on your deep learning PC. Simple, effective way to manage heat!
Creating a solid business plan doesn't have to be complicated. With Corvus Link, you can simplify the process and turn ideas into actionable steps to bring your startup idea to life.
Very insightful, the actually helpful kind of "a day in the life of" some kind of profession. Really helped me to get to know more about how an ML engineer works.
The best way to live is to Love what your doing. God bless you People...
Great video!!
I have a question.
Why build expensive PC's for model training when you can google colab or other cloud services to train your models?
I can think of several reasons:
1) Your own PC is just that: you can do whatever you want with it and whenever you want.
2) In case you didn't know: whatever runs on Google, stays on Google. So unless you want to share your company's trade secrets with Google, get your own PC.
3) The GPU's provided by the free version of Google Colab aren't that good and also get disconnected quite often. Buying the pro version gets you a better GPU and a limited number of computing units to use it. If you're running models 24/7 (as mentioned in the video), those units will run out very quickly so you'll have to buy more all the time. Best to just get your own setup and just deal with the electricity bill, which should already be handled by the solar panels these guys have on the house.
@@Jevgenij-z6x ohhh now I get it, tyy.
It looks so fun and inspirational. Love Huyen Chip as well
You are back! Wish you a good luck with your app. It's an awesome idea!
Reminds me of a very popular tv series, 5 years ago..with the prediction of LLM's ...btw very Aussie style with the food the exercises and the sun, man I miss the sun u get there,
Very nice :D I'm training my LLMs and watching pros in action. Good luck ;)