jq 'group_by(.dept) | map({dept: .[0].dept, total: map(.salary) | add})' data.json

jq is a JSON processor built around a pipeline of filters. Here group_by(.dept) buckets objects by department, then map builds one summary object per bucket, adding all salary values with add. The pipe symbol (|) chains filters, feeding each stage's output into the next. This is the jq equivalent of a SQL GROUP BY with SUM.