L'Oréal · Primly Community

L'Oréal senior / L5 system design interview: what to expect and how they evaluate

infra_ines · 4 replies

Just finished the loop for a senior software engineer role at L'Oréal's tech hub in NYC. The system design round was the most interesting part and I want to give it the writeup it deserves since basically nothing exists about this company's eng interviews.

First: L'Oréal calls their senior IC level "Senior Software Engineer" but depending on team it maps roughly to L5 at Google or E5 at Meta in terms of scope expectations. They're not using numeric levels externally.

The system design setup

60 minutes, one or two interviewers (I had two). No specific prep sheet was sent; the recruiter just said "be ready to design a scalable web system." Classic.

My actual prompt: design a platform for personalizing product recommendations for L'Oréal's e-commerce properties across different brands (Garnier, Maybelline, Lancome have different sites). Think multi-tenant, cross-brand user data.

What they focused on Data modeling: how do you represent a user across brands without violating data consent boundaries (GDPR came up immediately) Scale: they gave me rough numbers, something like 50M monthly active users across all properties Caching and latency: "customers expect recommendations to load fast on mobile" They did NOT go deep on distributed consensus, replication lag, or anything that felt like an infra exam. It was product-architecture, not infrastructure.

How they evaluate (my read)

They want to see you structure ambiguous problems, ask clarifying questions, and make tradeoffs with reasoning. One interviewer asked "what would you cut if you had to ship an MVP in 6 weeks" which told me they care about product judgment as much as technical depth.

I passed this round and got to the hiring manager stage. The behavioral interviewer on the same day was actually harder for me than the system design.

4 replies

director_dee

The GDPR angle on a multi-brand recommendation question is a great signal question. If you're doing a loop at a global CPG like L'Oréal and you design a user data system without mentioning consent or regional data residency once, you've already lost a point. Good on them for testing that.

ds_dmitri

Did the recommendation system question require any ML architecture discussion, or was it purely infra/backend? I'm a data scientist and wondering if L'Oréal's tech org has a separate ML/DS interview track or if it overlaps with SWE system design.

staff_steph

They asked about the ML component at a high level, but it was optional to go deep. I mentioned a collaborative filtering approach vs. a rules-based fallback for cold-start users and they seemed happy with that. I don't think they expected me to design the model training pipeline, more the serving and data layer. The ML team is probably separate.

qa_quinn

How much did they get into reliability and testing the recommendation system? Was there any discussion of what happens when recommendations fail or fall back to defaults?