I'll argue using what @1mdlrjcmed gives as a hint.
Suppose NOT, there is no such interval $A$.
Then for any interval $A,$
$$|f(x)| < \left(\frac{2}{3}\right)^n (n+1),$$
where $x \in A.$
Note that it implies that $M:=\max_{x \in [0,3]} |f(x)| < \left(\frac{2}{3}\right)^n (n+1).$
Also, observe that $$\int_0^3 \left(x-\frac{3}{2}\right)^n f(x) \, dx = \int_0^3 x^n f(x) \, dx= 3,$$
which is true since $\int_0^3 x^k f(x) \, dx =0$ for $k=0, 1,\dots, n-1$ and if we expand $(x-\frac{3}{2})^n$, then all the integrals of lower-degree terms $x^k f(x)$ equal to $0.$ So the translation(I mean the translation only restricted to $x^n$, not whole integrand.) won't make any difference to the integral.
Then
\begin{align*}
3 &= \Biggl|\int_0^3 \left(x-\frac{3}{2}\right)^n f(x) \, dx\Biggr| \\&
\le \int_0^3 |f(x)|\Biggl|\left(x-\frac{3}{2}\right)^n\Biggr| \, dx \\
&\le \int_0^3 M\Biggl|\left(x-\frac{3}{2}\right)^n\Biggr|\, dx\\
&\le 2 \int_{\frac{3}{2}}^3 M \biggl(x-\frac{3}{2}\biggr)^n \,dx = 3\,\frac{M}{n+1} \biggl(\frac{3}{2}\biggr)^n.
\end{align*}
Hence,
$${n+1}\biggl(\frac{2}{3}\biggr)^n \le M.$$
This contradicts to the assumption.
So the assertion follows.