hi I am reached out by MLE recruiter at Meta. They sent the preparing material: Being a data scientist I focus more on EDA, Understanding business requirement, building ML models and pipelines, AB testing, and some projects lead to production deployment. Honestly given the diversification of data science tasks, I dont have much time to focus on solving leetcode problem (I can do most of easy, and done 30-40 medium) for Python. Do you think the interview for someone with DS background would be similar to SWE? like asking medium/hard leetcode, system designs etc or it is combination of easy/.medium leetcode and stats/ML? If the interview track is same as SWE, I dont see a point going for a role which does not weight my experience and background (PhD in Stats, 5Yr YOE). I also have invite for Research data scientist and might focus to prepare for that role instead. Can you guys give some insights, if you work at Meta or have gone through the interview to enlighten me? #meta #machinelearningengineer #softwareengineer #datascientist #researchdatascientist
You will have multiple rounds of LC questions. Either go for DS or do LC. This is coming from experience for both DS and MLE interviews at meta. DS interview track has SQL interview not python
I won't go for DS as thats best suited for someone wanting to do lots of SQL and analytics. Something they have called research data scientist (RDS) I am more keen on it as if MLE is not mutual fit.
Based on Blind I thought Albertson has a high bar.
mle loop is swe but SD can be replaced with ML design. But why do you want to move from Albertsons? Isn't it highly regarded in blind community?
Is this Albertsons’ the grocery chain? I had no idea they kicked ass
WTH is EDA? Sorry, there's just TLA
Exploratory Data Analysis, feature engineering and data prep stuff when you spend most of time in model building process
With feature stores, that’s not how most models are built. It’s data prep, but mainly writing pipelines.
RDS is probably a better fit given the experience you describe. We do a mix of EDA, applied ML, etc. DM if you want to know more. MLE at Meta is much closer to SWE and interviews reflect that. Mostly leetcoding and ML design.
Sure, this is helpful. I will dm you
I did the MLE loop recently. First, it's basically the same as SWE loop except you have some ML round. Leetcode is a must. Just master the top 100 frequent FB tagged questions, and do Blind 75 List. Anyways I messed up the topological sort (Alien Dictionary) question so the recruiter offered to put me up for RDS role if I did another round. I am still waiting on the outcome from that..
Very helpful comment irrespective of which interview I take, thanks
Do MLEs get paid more than SWEs at meta?
I’ve done both RDS and MLE loops at Meta and it does sound like you’re more interested in RDS. But I just want to emphasize from your other comments that you will still need SQL and LC knowledge for RDS SQL isn’t just for product DS. In fact you’re gonna be tested on some product sense for RDS as well…
You definitely need some LC for RDS, but the bar is lower than for SWEs/MLEs. As for SQL, you can answer the wrangling questions with pandas/dplyr too. I agree about the product sense, but it's more about being able to attack a business problem with an e2e data science solution.
Is fb rds only for phd or do they take in masters as well
there's going to be LC. Take time to prepare. Grokking, Blind 75, FB tagged in leetcode(premium). I couldn't clear their phone screen . Study guide said no need to prepare for DP but I was asked dp(k char palindrome) and the 2nd question doesn't even exist in leetcode lol
MLE at meta is SWE with Ml knowledge. You aren’t fit for this role. Apply for product DS role
you decided faster than a recruiter? you a bot or something?
You want information or not? Or do I need to be recruiter to have this information