What candidates say the round looked like, summarised in our words, each linked to where it came from. Databricks hasn’t confirmed these details unless the source says “Official page”.
Had this round? Add your report →
Another solutions architect candidate had the same synthetic dataset scenario, but the interviewer kept digging into varying the grain of the data, generating it at scale and deep Spark details, so it felt more like a Spark test with GenAI data generation as the canvas. They got some visualisations out before time ran out. Their advice: brush up on Spark shuffles and query debugging more than any other part of the platform.
Solutions architect live vibe coding round. The guidelines discourage writing code yourself: the LLM generates nearly all of it and you explain it. The scenario: pick an industry, generate a few synthetic datasets, join them, compute some metrics and build a visualisation. The candidate first had ChatGPT draft a step-by-step spec, then ran each step in the Databricks Assistant, handling prerequisites like catalogs and packages first. Then the curve balls: scale one dataset from 10k to 100M rows, optimise the Spark code, deal with data skew. Know the Databricks UI before you go.