PwC · Primly Community

PwC data engineer interview, pipelines and SQL: sharing my full loop for the technology consulting practice

analyst_ana · 6 replies

just finished (and accepted) a PwC data engineer offer so here's the debrief. this was for their Technology Consulting practice, specifically a team doing data platform work for financial services clients. sharing because there wasn't much out there specifically on the DE interview, most posts conflate it with the DS role.

screening: standard call, 25 minutes. they wanted to know my pipeline experience (Airflow, dbt, Spark, that kind of thing) and my comfort with client-facing delivery. the word 'client' comes up a lot at PwC. get comfortable with it.

technical round 1: SQL and data modeling (75 min with a shared coding environment) this was the hardest round. two problems: first was a pure SQL problem. given a table of transactions with timestamps, compute rolling 7-day revenue per customer and flag customers where that metric dropped more than 20% week over week. window functions, CTEs, date math. not a "find duplicates" beginner problem. second was a data modeling question. they gave me a vague business description (insurance company tracking claims, adjusters, and outcomes) and asked me to design a dimensional model. star schema, slowly changing dimensions, grain definition. they pushed on every choice I made.

technical round 2: pipeline design + architecture (60 min) almost entirely conceptual. they described a scenario where a client has 14 source systems dumping flat files in different formats onto an SFTP server and they need a consolidated data warehouse refreshed daily. walk them through how you'd build it.

they wanted to hear about: ingestion layer (how you handle bad files, schema drift), transformation layer (dbt vs. stored procs and why), orchestration (Airflow scheduling, retry logic, alerting), and data quality checks. no coding here, just architecture discussion. they also asked how you'd handle a source going silent: do you fail the pipeline, do you load partial data, what do you tell the client.

behavioral round: standard STAR. biggest focus was on situations where a technical solution had to be communicated to non-technical stakeholders. PwC DE roles are consulting roles, you're presenting to clients.

partner call: 30 minutes, felt more like a coffee chat. they were checking if I'd be a good client-facing presence.

comp for the offer (senior associate level, NYC, 2026): base in the low 110s, which is below pure-tech-company DE comp but the benefits package and brand recognition for PE/banking clients is the trade-off people make.

tips: get comfortable with dbt specifically, it came up in both rounds. Databricks is in heavy use on their platform projects. cloud was AWS for this team.

6 replies

ux_uma

the SFTP scenario is so real. that's basically every legacy financial services client ever. 14 source systems is probably optimistic, I've seen worse. did they ask anything about how you'd handle PII in the pipeline or was data security not part of the discussion?

de_derek

they touched on it briefly in the architecture round, more as a 'what would you consider' question than a deep technical drill. I mentioned encryption at rest and in transit, access controls by role, masking in dev environments. they seemed satisfied but didn't go deep into GDPR/CCPA specifics. might depend on which practice area the team is serving.

market_realist

low 110s in NYC for a senior associate at Big 4. that's the trade-off in one number. how are you feeling about it vs your previous role?

de_derek

coming from a startup where the base was similar but equity was worthless and the team was 2 people, PwC actually felt like an upgrade on most axes except pure ceiling. we'll see in 18 months.

sre_sol

the 'source going silent' question is a great signal for maturity. a junior will say 'the pipeline fails and you get paged.' a senior will say 'depends on the SLA, here are three options, here are the trade-offs.' good that they're testing that.

sec_sasha

for anyone prepping: that SFTP+flat-file scenario also usually brings up file integrity questions that security cares about. checksums, signed manifests, detecting truncated files. if you can mention even a sentence on file-level validation they'll remember it.