The take-home assessment is a full day or 2 of work in 1-3 hours, with no allowances for LLMs to write syntax. I understand how an element of restriction on the latter is necessary to make sure the candidate is not "faking it" but I'm sure there are other avenues to do this. For example my assessment looked for: an optimized ETL job (Airflow DAG), well-documented SQL schema and dbt data model, and "bonus points" for Docker orchestration of all the above. Respectfully to just get ghosted I'm not sure who will do this let alone successfully while just googling the boilerplate code...