SnowPro Advanced: Data Engineer is the exam Snowflake engineers take after Core, and the current version is DEA-C02. Snowflake's page gives $375 per attempt and states plainly in its FAQ that each registration requires the full amount, so there is no discounted retake. The candidate profile asks for two or more years of hands-on experience as a data engineer in a production environment, and you need an active SnowPro Core certification before you can sit it. The format is widely reported as 65 questions in 115 minutes with a scaled pass mark of 750 out of 1000, but those numbers come from third-party sources; Snowflake puts the detail in its study guide rather than on the main page. The skills listed on the page are sourcing data from data lakes, APIs and on-premises systems, transforming, replicating and sharing data across cloud platforms, designing end-to-end near real-time streams, designing scalable compute for data engineering workloads, and evaluating performance metrics.
In practice that means a lot of questions about Snowpipe and Snowpipe Streaming, streams and tasks, dynamic tables, external and Iceberg tables, warehouse sizing and clustering. I rate the exam more highly than Core because the questions are built around scenarios. You get a pipeline with a latency requirement and a cost constraint and have to pick the right combination, which is a much better test than recalling which edition includes a feature. People who have only done courses tend to find it noticeably harder than Core, and that is a good sign. The cost is the obvious drawback. With Core at $175 and this at $375, the minimum is $550 before any retakes, and every retake is another $375.
The Databricks Data Engineer Professional is $200 and spends more of its weight on monitoring and cost optimisation, which I think are the most valuable skills in either platform. Snowflake does test performance evaluation here, but the cost side feels lighter for a platform where an oversized warehouse left running is the classic way to blow a budget. The other gap is AI. Snowflake now sells itself as the AI Data Cloud and ships Cortex functions, Cortex Search and Snowpark ML, but this exam is about classic data engineering. That is not a bad thing in itself, because good AI work still sits on good pipelines, but it means this cert says nothing about whether you can build retrieval or feature pipelines.
For that you would look at Snowflake's separate Gen AI specialty, which we cover elsewhere. Search interest is down about 30% on last year. I would read that less as the cert losing value and more as Snowflake's engineering market maturing. If you work in a Snowflake consultancy or partner, this is one of the credentials that actually gets noticed. Outside that world, a GitHub repo of a working pipeline will usually say more.