Вам нужно:
df = pd.DataFrame({'id':[1,2,3,4,5,6,7,8,9,10,11,12],
'level':['low','high','low','medium','medium','high','low','high','medium','high','medium','low'],
'values':[23,43,56,12,34,32,18,109,345,21,15,45]})
# use nlargest(10) for your problem.
print(df.groupby('level')['values'].nlargest(2))
Вывод:
level
high 7 109
1 43
low 2 56
11 45
medium 8 345
4 34