Groq to Meta?
I joined Groq (AI chip startup) about 1.5 years ago (based on US West Coast) as a compiler engineer out of uni and have had phenomenal growth here - lots of hard problems to work on, important projects to lead etc. (there is always good scope to find). Meta reached out recently about an interview, I did the full AI coding/ design/ behavourial loop and it went well. I got an IC4 offer for compiler-like projects related to PyTorch and was happy (standard per Levels, 345k USD). But, have a few questions I was confused on: At Groq, I've had the chance to lead important projects (as team is small) and enjoyed it. I see strong personal/career growth at Groq since I'm trusted by senior/VP-level folks and am involved in important design decisions but it is a long lead-time (still few years out until there is rapid growth or liquidity in compensation etc.). Comparatively, IC4 at Meta would double my liquid compensation (same base, but all stocks are liquid now). But, I'm scared I'm "resetting" my growth - whats the IC4 scope like at Meta and growth opportunities to more senior IC roles? Since PyTorch is extremely mature, is there still "project scope" to justify promotions? I love the pace of a company like Groq - I've been able to get exciting placement/memory algorithms prototyped and checked in in short experiment loops (since the team is small, collaborative). I've heard (never been at BigTech) that there are more processes/systems such that pace at Meta would be slower? Whats the "politics/expectations" at IC4 - can I find interesting problems or is it mainly TL/manager-driven? Thank you! Current TC: 175k + 65k in Groq stocks/yr at Series C val #meta #tech #compiler #softwareengineering
How long is the Meta offer good for?
I got it last week and was given 4 weeks so about 3 more
I don't know what motivates you, but it seems like you're happy at Groq. Everything you said about Meta is likely true to a degree, as bigtech is exactly as you described depending on the team or org.., but you have negotiating power in that you don't need to move unless it's a great deal. The inertia to overcome your move to Meta (or any company) should be reflected in either the job (being more awesome than what you're doing) or in pay (substantially more money to make it worth your while and risk) or possibly in long term growth prospects..
thanks!
All your fears are true. And what's worse, the MITA accelerator may get canceled. Meta are really only building it to threaten Nvidia into giving discounts. The MITA accelerator is...... not good. Groq is actually innovating in the space, MITA is one of half a dozen toy AI accelerators from large tech companies that's going to go nowhere (Google TPU and AWS trainium will win, because the have huge cloud infrastructure already backing it). You'll have to prove yourself first at Meta (1 year) before you get given anything cool to own. And you'll have to compete to get it. There are numerous old timers at Meta that hoover up the opportunity and you'd be a day 1 employee. Honestly, stay at Groq. Or Google TPU, Nvidia, AWS, AMD, .
I don't think this role is within MITA as far as I understand, more like an internal-facing team building additional compiler infra for OSS PyTorch. What is the "timeline" like at places like Meta? aka "1 year" at Groq meant lots of scope growth (I can now own parts of the compiler). What can be expected average-case/best-case for the kind of influence/ownership "1 year" at Meta?
At AWS we pay senior engineers in the Trainium compiler like 700k a year. That's thr kind of experience you are getting right now at Groq. It's not just AWS - google, nvidia, pay top dollar for solid ML compiler experience. Groq is well known, it's run by some extremely sharp folks. Intending on actually making money by selling chips. Meta is using AI mostly to consolidate advertising revenue. They don't sell anything expect ad space and a AR headset. If it's just internal facing tools and extensions to PyTorch you'll be missing out. If you want a reference for a real compiler team at Amazon, reach out. But honestly, take the offer to your manager and just explain hey - I love it here but this money that Meta is offering is important. It buys houses and sets up my future. Can you match it? I understand if not.
Show meta offer at groq , bump it up cash wise or get more shares and stay at Groq
Yup that's the way.
"there are more processes/systems such that pace at Meta would be slower" this is 100% true. Big companies expect less technical work and more showcased work, as you are related to compilers showcasing will be tough. Check the senior folks in current company growth if management has history to give growth stay else jump.
I've been "informally promised" the opportunity to lead a small team about 8-10 months out (based on internal roadmap that I can't reveal too much).
Stay Groq bro. That's a cool opportunity. I'd love to work at Groq, work at AWS now on Trainium. I never heard back :-(
You are at the beginning of your career, and Groq is highly likely to get acquired by aws/azure/gcp so I would suggest stay put and then ๐ธ๐ธ๐ธ๐ธ
Groq is a Llama inference monster, llama being made by meta, it's kinda possible OP ends up working for Meta at vastly higher pay if Meta acquires lol
You are losing your once in a lifetime chance to be a millionaire by leaving Groq People are killing themselves to join Groq
Stay at Groq. In Meta you would be just another engineer. In Groq you have chance to make history of AI chips etc. and if you can refer me at Groq that would be great.
Stay. Many Meta engineers are doing very boring work
does groq give refreshers? I'd stay if they do
I got a base raise (performance driven) at the end of my 1 year to get to 175k, not equity refreshers. Could discuss with firm