Try it

df.pivot('name', 'subject', 'score').plot.bar()

The basic pandas visualization principle is to use pivot or melt to create the table format required for charting.

The following is the code I ran (Jupyter Notebook)

% matplotlib inline
data = [ {"name":"abc", "subject":"A", "score":40}, {"name":"abc", "subject":"B", "score":60}, {"name":"abc", "subject":"C", "score":40},
        {"name":"xyz", "subject":"A", "score":10}, {"name":"xyz", "subject":"B", "score":90}, {"name":"xyz", "subject":"C", "score":30}]
df = pd.DataFrame(data)
df.pivot('name', 'subject', 'score').plot.bar()

Result: