0. Trolls go away, come again another day. 1. Want to build something using ML. I am stupid AF and can't read books or watch theoretical explanations. I only learn step by step and when I need to learn.. 2. For amazonians, NO i ain't going to MLU. I am pip'd and now unemployed and alone. 3. Andrew NG free course is too theoretical for me. I tried it, couldn't go past 2 lectures. 4. Is there a github handbook to follow? I can clone and reverse engineer it to learn. I think I can learn that way. 5. Again I am stupid AF, so yeah I know I should give up, but I still wanna waste some energy/time for the umpteenth time. 6. Please no long pdfs with mathematics, I just want to build something - I will learn how it works internally when I need to. 7. CV , NLP, whatever seems easy I want to try. I want to try it all. #amazon #twitter #netflix #salesforce #apple #Google #korea #india #usa #blind #pip #uber #facebook #unemployed #japan #australia #machinelearning #machinelearningengineer
If you want someone relatively easy, stay as swe. ML is even harder.
Project idea-Visual search recommendation system
could you elaborate. Is there an existing github project I can look at?
There should be. For a basic idea of architecture look at the Pinterest visual recommendation system on youtube.
Go back to Amazon and tell them to add automated subtitles on twitch
They kicked me out and burned the bridge. They won't take me back. The only way to fix twitch subtitles is if YOU go in and save the day.
For most DS models, the data used is proprietary so you do not have many options to build a product due to lack of data. To overcome this, you can either do DS consulting for data rich companies that are struggling to build models or start collecting data and build generic models that can be used by different companies that do not have in-house data. Another option is to explore uncharted territories where DS can be used next. It’s being used in health care, credit decisioning, marketing, HR, manufacturing etc. So try finding another niche industry where it’s not being used currently and develop products accordingly
Good advice. At this pont in time, I don't understand what the model is. Types of it, uses of it etc..
Do you have any DS/ML experience previously or you are trying to learn and then develop a product?
Try aws's sagemaker templates. There are basic ones. you can build an endpoint and deploy it on a lambda. That's as interesting as it gets without understanding math behind it
I will give it a shot
Yes it's part of the free tier, either way just set up a limit spend of 5 a month. Actually I forgot to turn off a service for 4 months and they refunded me. It wasn't much anyway
🤔 thanks
If you like NLP, huggingface transformers have a great hands-on course. You can then fine-tune some great state of the art models.
Where can I find this course?.Is this free?
Yup. There you go: https://huggingface.co/course/chapter1
I found kaggle ML course a good high level intro and practical. Give it a shot.
will check it out. Didn't know kaggle had courses too. Did you have prior DS/ML experience?
Try building a movie recommender system by using movie lens data set. I am surprised to see a lot of retails companies don't even have a decent recommendation feature on their website. Definitely a huge untapped market. Once you understand the basic it can be applied in any domain.
i have tried this in the past as a college project. Not much fun 😶
@OP you did? What maths did you need to know for it? Why wasn't it fun?
Interesting hashtags choice
Maximizing the reach!