Apple vs. Nvidia?
I am a new grad with offers at both nvidia and apple. They’re both similar in TC but nvidia has a higher base salary with less sign on/RSU. The apple position is working on iOS/MacOS kernel and user programs and Nvidia is on the deep learning library team. Which would be better to take if my end goal is to move to the east coast and work in NYC?
Yeah.. TC? Choose the team for your growth (don’t think abt location now)
NVidia should be more exciting and educating, Apple should be more $$
Non tech roles?
Apple has great refreshers and ESPP. So your TC might be more at Apple even with the higher base at nvidia. Which team do you like more?
It’s hard to say, I like deep learning/machine learning but I feel like the nvidia role could be mostly performance analysis rather than implementation. Apple will be a very multidisciplinary role between firmware/software/algorithms so I will get to learn a wide range of skills and explore my interests. I’ve also interned at Apple so feel a bit more comfortable there.
As a fresh grad - choose the offer which gives you the most learning and sets you up for the future. May sound like old man advice - but give yourself a few years learning the most in demand skills and look at how your TC and career takes a turn for the good. Unless you really have a pressing need for the extra money, choose learning and skill growth over money while beginning your career. There will be a time when you might NEED a higher TC, then you can choose the other one
Yeah, the TC is insane for both for a new grad in my opinion so I have no problems there. I think the team at Apple will be a lot of fun to work on and I’d learn a wider range of skills and have a lot more responsibility. However, nvidia will let me learn more about deep learning on a lower level which is very hot right now. I’m a bit concerned it’ll be more performance analysis than implementation though (not sure if this is a big deal). What do you think?
It's a tough choice and it's great that you are having this problem 😀 Deep learning is a very valuable skill to have for the foreseeable future. I will probably lean slightly towards that but can't underestimate the power of learning a broad set of skills early on in your career. Maybe you can do that too at NVIDIA along with deep learning expertise?
Ha ha
Deep learning, at a lower level, is number crunching. If you are working on a lower level it is a high performance computing, squeezing most out of Hw and suggesting changes to HW. Still an interesting problem, but you might want to understand exactly what kind of problems you will be solving. (Ask ur hiring mgr what is the biggest problem he has right now) One good thing about NV is flexibility. Out TC isnt as good as others but we are more flexible with remotes. I was once in a team of 6, every single eng and mgr in a diff location
Remote is a huge plus, but to your first point, are you saying that doing deep learning library performance stuff isn’t going to necessarily help you break into the field of deep learning/machine learning since the primary focus is on high performance computing and not understanding the models themselves?
Deep learning is a huge field and a stack. TSMC is a part of the deep learning stack. Bits and flops are a part of the stack. If you are interested in deep learning lower level stack, you will love it. But if you are interested in using deep learning to crack the application side of deep learning, just working on deep learning library probably wont suffice. Most likely you are trying to figure out how to maximize the use if HW for a given SW deep learning function. The application side of things are more if data science But once again, having that knowledge in lower level is important and useful. It is more of a question of what you want to work on.
Nvidia tends to be generous with remote work policies. Many people in the DL software team(s) work remotely, although you may have to spend a year or two at HQ or a branch office before they’ll let you go remote.
The position is deep learning library performance SWE. Will I gain experience that will be valuable if I want to transition into ML and do more algorithms stuff? It seems like this position is a bit lower level.
Pros: You’ll learn a lot more about the hardware, the math, and the full software/hardware stack than you would doing pure ML. This will make you hideously marketable. Not a lot of ML researchers have that full stack knowledge. Cons: If ML research is where your heart lies, this role isn’t exactly it. Might not be worth spending a year or two of your life. Personally, I think performance optimization is more fun than data wrangling and running experiments ;) Fyi nvidia permits internal transfers although you will still have to interview with the destination team (policy varies from team to team, typically not as much work as an external interview but they will ask some honest questions)
OS kernel type of job is unlikely to be found in NYC.
The nvidia position is more focused on DL library performance on the lower level. Is that going to be just as hard to transition into a NYC job? And does what I learn have to be directly applicable to find a job in NYC? I’ll be learning great software engineering practices and C++ in depth.
Maybe you can work on high frequency trading... They probably use C++.
What is the offer?
130k/50k/30k-Apple 145k/40k/20k-nvidia base/RSU/sign on
RSU for 4 yrs?