If you want to self study ML without wasting time here is the definitive guide since I've gotten so many DMs on this topic: Just answer this one question: - do you want to be a cracked researcher or a rlly fucking good engineer? TLDR; 1) engineer path? read PyTorch docs, a book, or a YouTube tutorial on it -> watch ANY build from scratch video -> read and implement papers from scratch until you can do it without much help -> you're likely also good at coding apps or websites (I loved apps first then websites) so start making real apps for fun or for money with the shit you've learned -> dissect how big real world companies are designing their ML systems (blog posts usually) - the trick here: GO WIDE. Get good at multiple aspects of ML engineering—date engineering with spark, building models, fine tuning models, design CI/CD pipelines with the data orchestration (idk use airflow or stepfunction or something), understand the deployment process in production (MLOPS), dive into scalable architectures from Pinterest, Google, META, etc. 2) research path? learn the **geometric intuition** of calculus and linear alg -> learn probability theory rlly paying attention to the notation since you'll see a lot of that in papers -> learn the intuition of classic trade offs in statistics like bias/variance, what happens when your dataset is noisy, etc. -> learn classical ML theory using ANY textbook or Andrew Ng or a uni course-> learn the early DL works before reading Attention is All you Need -> CNNs, NLP, LLMs, IT DOESN'T MATTER, choose whatever is cool to you in a given moment and just understand papers from a intuition perspective like "ah I see why they would add a final sigmoid activation at this step since we're working with probabilities for multiple classes" or "ah I see why they actually need attention here because vanilla RNNs would just lose that context in its hidden states" etc. -> repeat with X amount of topics within DL - the trick here: GO DEEP. Intuition. Intuition. Intuition. This is what I lack when reading research papers lately cuz I'm so optimized for engineering but if you wanna be a rock start researcher get that INTUITION! Okay that's it you can stop reading but here's info about me: I opted for the research path for about 2 years but now I'm only optimizing to be the best ML engineer I can be. It's just more fun to me at the moment and early on I needed money ASAP so was not ok with doing the PhD route. I'm okay with my theory knowledge not always being 100% there since I'd rather be able to build and design systems better than most others in a room I keep getting questions on "what resources" I recommend for self studying ML and the truth is it doesn't matter. There are thousands of amazing resources out there and tbh you can probably choose any random one and get really good results as long as you know what you want. If I really had to recommend them though, just look at my profile I post a lot of gems. Notable ones are: - ISL and ESL stats books - Andrew Ng's coursera - 100 pg ML book - Designing Machine Learning Systems book - Pinterest ML blog on medium - data engineering Zoomcamp for spark - math academy for all your math needs - Karpathy's zero to hero stuff You gotta realize that there's limited time in a day and if you're juggling ur day job, the gym, family, maybe even a child, then you better choose the right things to work on to save yourself some time. Go read the TLDR; above again and choose your adventure! It's fun up ahead if you choose the right path for you

@acyanesz Jus hit 4 phew

@soysarcasme 🔒

Office dawg looks so human

@kuberdenis Let’s go!

@shurensha Do you feel like you retain all the info?

I was catching up with @mukulsurajiwale a other day and he mentioned how it kind of doesn't make sense for AI engineers to give daily stand up updates since ML progress is highly non linear If you're working in the space how do you think progress should be reported?

@jxnlco bro just leaves the hardest quotes and leaves me to ponder for a few months appreciate you

@_cat_turner im busy all day but it's those damn nights before bed

@dylandavids0n you're right :/

@jxnlco :'(

@jaegpark need to get on that

@verrsane time to lock in

@syn_nxs nah fr you right... I NEED TO LCOK IN

@ivanleomk since oct

@dev_maims Designing the system / planning: 1 week

Just played basketball until 1am lmfao wtf

it's gonna sound cringe as fuck but i felt so good today when i heard from a senior applied scientist on my team that my manager regarded me as a very strong MLE that she could work with on this new project we are designing right now tbh i feel imposter syndrome every day still im surrounded by the smartest people I've ever worked with on my current team they make me feel so stupid sometimes and i often wonder if im performing well enough but small things like this keep me going and makes the grind worth it the hours outside of work, learning from you guys here on X, falling in love with the craft... it's all paying off we all got this <3

@novasarc01 Senpai

@__berb IM A MENACE