J'ai postulé en ligne. Le processus a pris 5 semaines. J'ai passé un entretien chez Meta (New York, NY) en sept. 2025
Entretien
First round was a recruiter screen. The recruiter told about the role and expected salary. The next round was technical screen with a data scientist. He asked SQL and A/B testing questions.
Questions d'entretien [1]
Question 1
The interviewer asked two SQL questions that had to be answered on CoderPad followed by a product improvement question. I said that we need to do A/B testing and provided all details needed to setup the experiment. The interviewer seemed satisfied and was able to cover all questions he planned to ask. He had to nudge me a bit both for SQL and A/B testing questions but I was able to pick up fast and responded appropriately. I had a positive experience but received a rejection next day.
J'ai passé un entretien chez Meta (Menlo Park, CA)
Entretien
Started with a recruiter/screening round and then moved into the full loop. The full loop had four rounds focused on analytical execution, analytical reasoning, technical skills, and behavioral. SQL was a major part of the process, along with product metrics, experiments, stats, and communication. The process was structured but still challenging because the questions were open-ended and required clear thinking, not just memorized answers.
Recruiter screening then 1 hr technical screen on sql and business case. The interviews were focused heavily on product thinking: defining clear success metrics, interpreting metric trade-offs, and designing solid experiments. The SQL portion was straightforward but expected clean logic and structured thinking. The case rounds really test how well you connect metrics to real business impact.
I got a little caught off guard because I was expecting only 25% coding but it was like 75% SQL coding (5-6 questions related to a business case) followed by a full research design of an A/B experiment. All of this was expected to be complete in 40 minutes so you need to be able to work really fast.
Questions d'entretien [1]
Question 1
One SQL problem required calculating correlation from scratch using SQL, was not expecting that.