So there're multiple answers possible thanks to Ben Grossmann and user19077881, let me summarize it here (I'm new so I can't upvote Ben Grossmann's answer):

row = df.query("t == 3").astype(object).squeeze()
row = df.loc[df["t"] == 3].astype(object).squeeze()

The order is important:

1. Convert the whole result of the query to "object" so pandas will not convert ints to floats and keep a global "object" type
2. Use squeeze() to transform the queried row to a Series

And if there are multiple rows matching the query then extract the wanted row by using iloc between astype and squeeze:

row = df.query("t == 3").astype(object).iloc[wanted_idx].squeeze()