Thats right. I'm trying to find a way into traditional backend engineering. ML is full of charlatans who are trying to sell overengineered black boxes to each other and I want out. My language of choice is Python but I don't think that's the lingua franca for actual backend engineers. I've been doing this for 7 years now. How do I get out of this hole I've dug myself into? TC 200k
ML for sure overhypes and underdelivers but I see this more on the business side. I’ve been in several companies where eng are doing a great job (yes there are bad ones who are just here for hype and quick gain, but they are minority) to me it’s exciting because it is frontier… I build ML tools and there are whole big areas completely underserved today
Agree to this Mind sharing your level and yoe?
Senior 9 YoE
dude, stay in MLE if you want to do easy work, have good wlb, and also the coolness and demand. SWE is real work and you can't fool around like in MLE. Why do complex mentally demanding work for the same pay ?
lol MLE is a joke compared to SWE. There is a reason why every Tom Dick and Harry who doesn't even know CS wants to do ML.
My friends in mle mostly working on ml based projects and don't use any ide, no strict code reviews and pretty chill. Mostly is building models and reading papers instead actual swe activities
Same boat. Switching from ML to ML Infra/Backend Infra - much more exciting than training similar god damn models or improving data
That sounds like a good plan. How does one move into ML infra positions?
I would say look up MlOps, etc. Another way to know what’s useful is go find ML Infra/Platform roles and see their description - technologies they use. Read about things like MichelAngelo, MlFlow, Ray, Airflow, etc
Isn't it easy, good wlb, and more pay?
Not sure of the more pay part, think its similar to SWE. I'm just seeing a lot of hype-driven sleazy kool-aid when it comes to ML. I'm wondering if that's the case in the rest of tech or if its just ML
MLE has significantly more pay in companies like Apple, Linkedin and Amazon (Applied Scientist). Maybe Pinterest too. In other places, its similar to SWE.
What does the mle research involve? Is it just reading papers to try to find which new arch scores the highest on some benchmark that is similar to what you’re running in prod?
Learn Java Spring
😂
Expedia, I am not understanding what is so funny. Java Spring is very commonly used for backend micro services.
I’m on a similar boat. I have started reading papers related to infrastructure and when I compare them to ML papers, the latter seem to be written to confuse the reader.
can you please tell me where (what conference) can I find infra papers at?
Can definitely emphathise with ML papers being complicated just for complications sake.