DE AI-Native Full-Stack Round

Single 60-min Data Engineer round threading Business Case + Data Modeling + SQL + Python through one continuous scenario. AI is allowed in most sections but disallowed for SQL. Requirements One 60-minute CoderPad session, one unified business scenario, four back-to-back sections with no breaks. Section 1 — Business Case (~15 min, Plain Text). Interviewer pastes a product-scenario paragraph; you respond in writing. Two typical sub-parts: (a) clarifying questions, (b) translating business goals into technical terms. Evaluated on: spotting ambiguity, multi-stakeholder thinking, awareness of both success metrics and potential harms. Section 2 — Data Modeling (~15 min, Plain Text). Sketch fact/dim tables, primary keys, partitioning, grain. Free-text, no SQL yet. Builds on Section 1's resolved requirements. **Section 3 — SQL Debug (~15 min, PostgreSQL 12.4, runnable). AI explicitly disallowed in this section. You're given a broken query against the schema from Section 2 with multiple bugs (4+ reported). Find and fix them unaided. Section 4 — Python (~15 min, Python Project mode, runnable). Implement a transformation / aggregation on top of the data from Section 3. AI re-enabled. Notes The four sections are tightly coupled — bad clarifying questions in Section 1 propagate into schema mistakes in Section 2. The SQL section's no-AI rule is the single most important constraint candidates miss. Practice spotting bugs (joins, group-bys, nulls, window-fn frames) under a 15-min timer. Level: IC5 / IC6 Data Engineer. Preparation Run a self-test: pick a moderately broken 30-line SQL query (Stack Overflow examples work) and time yourself to 15 min, no AI. Practice clarifying-question banks (volume, latency, data sources, stakeholders) so Section 1 reflexes are instant. Drill star/snowflake modeling on 3-5 product scenarios. Read the full debrief thread end-to-end — it's the most complete map of this round currently in the community.