Salesforce · Primly Community

Salesforce data engineer interview, pipelines and SQL: full loop breakdown 2026

de_derek · 4 replies

did the Salesforce DE loop a couple months ago for a mid-level role on the data infrastructure side. they have multiple teams doing DE work (MuleSoft, Tableau integration, core CRM analytics) so your loop might vary depending on the team. mine was for core CRM.

phone screen - 30 min coding in CoderPad. one SQL question: given a salesactivity table with timestamps, find reps who had at least 3 activities in any rolling 7-day window. window functions are your friend. they want clean code, not brute force.

technical deep-dive - 90 min. this is the meaty one. three sections: system design. they asked me to design a pipeline that ingests CRM event data (think: account updates, opportunity stage changes) in near real-time and makes it available to downstream reporting. talked through Kafka for ingestion, Spark Structured Streaming or Flink for transformation, Delta Lake or Iceberg for storage, dbt for modeling. they cared a lot about how i'd handle late-arriving events and schema evolution. SQL problem. more complex than the phone screen. involved multiple joins across a star schema (factopportunities, dimaccounts, dimreps) and some window function logic to compute attainment % by quarter. if you're rusty on dimensional modeling, brush up. debugging scenario. they gave me a fake broken Airflow DAG (shown as pseudocode) and asked me to find the issue. it was an idempotency bug: the task was re-running and double-inserting records. talked through upsert patterns and how to make the DAG safe to retry.

behavioral - standard. they asked about a time i pushed back on requirements that would have created tech debt. and a time i delivered something under tight constraints. Salesforce uses a 'Ohana' culture framing in some of their behavioral stuff, so expect questions about team and trust.

hiring manager chat - not an interview exactly, more a conversation about the roadmap and fit. asked thoughtful questions, felt more like a sell.

the whole thing took about 4 weeks. i got the offer. one thing i'd say: if you're coming from a pure SQL/ETL background without streaming experience, spend time on Spark Streaming concepts. they're genuinely building in that space and they'll probe it.

4 replies

sdr_sky

the idempotency debugging question is such a classic DE gotcha. upserts vs. inserts, making DAGs re-runnable. i've seen teams ship broken pipelines because nobody thought about what happens on retry. good signal to include in an interview honestly.

content_cole

the star schema SQL question shows up in a few team loops there. dimensional modeling is one of those things DE candidates sometimes skip assuming it's 'analyst stuff.' it's not.

ops_omar

thanks for the MuleSoft note. i'm interviewing for a DE role there and wondering if the loop is totally different. did anyone talk to you about which team it was during the recruiter call or did you find out later?

de_derek

recruiter told me the team in the first call. definitely ask early. MuleSoft DE is more integration/API-pipeline flavored vs. what i did. probably more emphasis on REST, event-driven patterns, maybe less Spark.