SQL fundamentals + correctness Joins (inner/left), window functions, deduping, aggregation, handling nulls “Latest record per key”, sessionization, retention-style queries Performance thinking: partition pruning, avoiding exploding joins, using appropriate filters Python fundamentals Writing clean functions, parsing/transforming data, error handling Working with dates, JSON, config-driven pipelines Unit-test mindset (what would you test, edge cases) PySpark / distributed data DataFrame API vs SQL, when to use each Joins (broadcast vs shuffle), repartition/coalesce, caching, skew awareness Window functions in Spark, incremental processing, idempotency File formats (parquet), partitioning strategies, reading/writing S3/HDFS

My experience interviewing with Autodesk has been extremely disappointing. This is the second time I have gone through their full interview process, only to be rejected at the very end. In both cases, I received positive feedback throughout the process and was advanced from one round to the next based on my performance. The interview loop consisted of four rounds and required a significant investment of time, preparation, and energy. What makes the experience particularly frustrating is receiving only a generic rejection after completing the entire process, despite being encouraged throughout the interviews. While I understand that not every candidate can be selected, a more transparent and respectful process would go a long way. Based on my experience, candidates should be prepared for a lengthy interview process and manage their expectations accordingly. Unfortunately, I felt that a substantial amount of my time was invested without meaningful feedback or clear communication regarding the final decision.