Tech Industry
Yesterday
638
Database companies that pay well for Staff SWE
Health & Wellness
10h
1314
Farting more in 30s. Is it normal?
India
3h
468
What do vegetarian Indians eat for protein?
Tech Industry
Yesterday
3085
1 vs 5 Million - no lifestyle change
Ask Blinders
Yesterday
800
Why is our country owned by Israel? I don't want my tax dollars fund genocide. How can we stop this nonsense?
Hi Community, I am prepping for the onsite FB rounds for MLE and the recruiter told me there will be a SDI round. I am going over some of the resources he shared and others available on the web but it all seems a bit overwhelming. For example, in designing a Dropbox exercise there are many protocols you need to remember and come up with new ones too. Can someone who has been through MLE FB SDI interviews comment on the difficulty of the interview? Do I need to prep topics like internet protocols, micro-services architectures, messaging queue architectures, client side programming, non-RDBS DBs, etc? If I do need to, what is good resource to start and how much depth should I dive in? I would appreciate any pointers. #machinelearningengineer
I wasn’t explicitly asked those things (had interviews 3 weeks ago) and was told to focus on a big picture so that I could create the whole solution without burying myself into details. Only once I mentioned something more detailed (potential cases when something in a queue can go wrong; it was based on one of the pictures from “Designing intensive...”) and the interviewer was super happy to hear that. Probably it gave me a pass
Hey can I dm you pls?
Please do
check out mlengineer.io
I went through this a few years ago and it’s the one interview I bombed at my MLE interview round at Fb :( — I prepared ML aspects of SDI in general (recsys, search/ranking, graph systems, etc) but the interviewer was clearly uninterested about that part, and all they wanted was to deep dive into seemingly random and dark concepts of Data Bases (particularly scalability), which I was clearly not prepared. A quick query just led me to great material like https://github.com/donnemartin/system-design-primer or https://github.com/checkcheckzz/system-design-interview and there must be much more out there that should help you, but yeah you should study and expect any of the topics you listed to be asked. I hope this helps! Fwiw this was early 2019 and now they could have change their focus/scope for MLE SDI.
They now have an explicit split between system design and ML design. System design is just that, no different from SWE system design, while ML design focuses specifically on ranking/recommendation systems (you need to talk about scalability in general terms but don't need to go into detail about the specific technologies used). Expect at least one of each type of interview.