Citi · Primly Community

Citi data engineer interview: pipelines and SQL, what they actually asked me

de_derek · 6 replies

okay so i just wrapped up a Citi data engineering interview for their enterprise data and analytics group. here's the honest breakdown, because the prep resources out there are mostly generic "prepare for SQL" advice that tells you nothing specific.

five rounds total. recruiter screen, then four technical/behavioral rounds over two virtual days.

the SQL round was heavier than i expected. not just writing a query. they gave me a schema for a financial transaction system and asked me to identify performance bottlenecks in a slow query. then write an optimized version. we talked about indexing strategies, partitioning, and query plan reading. this is not LeetCode SQL. this is "you've maintained a warehouse before" SQL. if you've been a DE for a few years you'll be fine. if your SQL is mostly analyst-level SELECT statements you'll struggle.

pipeline design. this was my favorite round actually. classic system design for data: design a pipeline to ingest real-time transaction events, aggregate by customer, handle late arrivals, and sink to a data warehouse. i talked through Kafka for the stream, spark structured streaming for processing, delta lake for the sink. they asked about exactly-once semantics and i went deep on it. know the tradeoffs between spark, flink, kafka streams at a high level. they didn't care which tool i chose, they cared that i could justify it.

coding. one medium-difficulty python problem. nothing crazy. data transformation, some string parsing. the kind of thing you'd actually write on the job. no dp, no graphs. i think they just want to know you can code.

infrastructure and ops. this one surprised me. they asked about orchestration (airflow vs prefect, i talked through both), monitoring pipelines in prod, alerting on data quality. there was a question about what i'd do if a pipeline started dropping messages silently. i told them my war story from a previous job. they loved it.

behavioral. straightforward. tell me about a time you had to rebuild something from scratch. tell me about a time you disagreed with a stakeholder. STAR format, no curveballs.

the whole thing took about 5 weeks. the team i was talking to was working on regulatory reporting pipelines which explains why reliability and auditability were such recurring themes.

my offer: i won't give the exact number but for a senior DE level in NYC in 2026, it was in the range you'd expect for a large financial institution. not FAANG, but solid and benefits are genuinely good.

happy to answer questions.

6 replies

backend_bekah

the 'silent message dropping' question is such a good tell for whether you've actually run prod pipelines. bet a lot of candidates who haven't had that experience bombed it.

infra_ines

did the infrastructure round have anything on kubernetes or containerization? or was it mostly orchestration/observability focused?

de_derek

mostly orchestration and observability. they mentioned their stack uses some k8s but didn't ask me to design anything container-related. more about how you think about pipeline health than infrastructure topology.

quietquit_quincy

was it all virtual or did they bring you onsite at all? trying to figure out if this is a 'you'll be in the office' kind of role

de_derek

all virtual for interviews. the role itself is hybrid, 3 days in NYC office. they were upfront about it. not fully remote.

visa_vik

did they ask anything about sponsorship at any point in the process? i'm on H1B and always nervous about bringing it up too late