Tech Industry
Yesterday
2237
The end of Backdoor Roth?!
Working Parents
Yesterday
934
What do you think is wrong with a kid who got rejected by 9 colleges?
India
Yesterday
986
Modi is a legend, will be remembered for centuries to come
Working Parents
13h
1075
Closed now - thank you all
Tech Industry
Yesterday
2814
Quitting this Slave life
I joined Google because of the prestige (google comes at the top of every single prestige poll on blind), the TC, and the opportunity to potentially transfer to an ML team. But I’m not really super passionate about all the infrastructure work and all the tooling work and that’s the kind of team I’m on. I feel like there’s already a saturation of “best practices” for infrastructure and software engineering, and you’re just following those and there’s not much room for creativity. Is this true? Anyone else feel the same way and wish they were working on some exciting ML or research projects? TC 200k
Lol!!!!
did you really make a decision based on blind polls? Everyone wants to work on ML - few are qualified
No I knew it was the most prestigious and I joined because they said there’s an opportunity to switch to research/ML with 20% projects and team transfer after a year. I’m just curious if anyone else feels the same way as me
I am qualified to do ML
Honestly I don't think work is ever all that interesting no matter what you do. Even when I'm working on my own projects there are huge swaths of work that you just have to grind out if you want to finish it. I wouldn't put too much stock in finding a job that enthralls you 24/7. Just focus on TC and learn to enjoy the grind.
If you want to do creative work, start painting. Building software is hard. It is about infrastructure, it is about writing tests, it is about taking care of versioning, it is about writing good CLs, it is about managing state. Even if you work for startups, it is about taking care of your customer needs. Even if you work in an M/L team, it is about Data Cleaning. It is about debugging. It is about performance, it is about explain-ability So, first you reset your expectations of your role in this industry. You are paid to the hard, boring things. If you really want to be creative, make those hard, boring things easier for others. There is no magical la-la land where you just sit and think ideas and your ideas will be perfectly delivered to the end user by a factory of minions.
💯
Damn it, you just repeated what I said, but better... 😥
Xoogler here. What I observed is, people burn out in initial 3 years and then go to comfort zone
:(
Burn out how? I thought Google was chill?
Apply to Flexport, ML is ground floor but problem Flexport is solving is arguably biggest TAM in world and interesting beyond just tech
I’d suggest chatting with folks who actually work in ML at Google. It might not be as sexy as you think. Many “ml engineers” are basically just tuning hyperparameters on models they have no control over for 8 hours a day.
ML Engineer is a fancy word for hiring SDE’s to do DevOps work on behalf of Applied and Research scientists.
Interesting. Not the case from what I've seen at Google & Waymo.
ML 1. Get into office 2. Run borg jobs that optimize hyperparameters 3. Write down results you receive from jobs from the previous day, usually by email 4. Be a disorganized shit like me and forget where you wrote certain things you are looking for 5. While waiting for your jobs to finish, write python scripts to automate some of your workflow or hack infra code to support your models. 6. Good luck showing "impact" to promo committee because you had like 2 merges to production in the past 5 months. 7. Switch companies and become a SWE
Haha yes! Except impact isn't always hard to show, depending on team. There are also other things: write rater templates, manage ratings, look at lots of examples (write up spreadsheets of annotations), work with analysts, some feature engineering, some math, etc. But most days are like you say.
Welcome to google... wait I don’t work there wtf am I saying