SKETCH: First prove existence, then worry about uniqueness. Clearly for $n=1$ we may take $k=0$ and $a_0=1$. Now let $n>1$, and suppose as your induction hypothesis that every $\ell\in\Bbb N$ with $\ell<n$ has an $m$-adic representation. There is a largest integer $k$ such that $m^k\le n$.
- What is an $m$-adic representation of $n$ if $n=m^k$?
If $n>m^k$, there is a largest integer $a$ such that $am^k\le n$.
- Show that $1\le a<m$.
- If $n=am^k$, what is an $m$-adic representation of $n$?
- If $n>am^k$, let $r=n-am^k$, apply the induction hypothesis to $r$, and get an $m$-adic representation of $n$.
For uniqueness, suppose that $n$ is the smallest positive integer with two $m$-adic representation, say
$$n=a_km^k+\ldots+a_1m+a_0=b_\ell m^\ell+\ldots+b_1m+b_0\,.\tag{1}$$
Show that $a_km^k+\ldots+a_1m+a_0<m^{k+1}$, and use this to show that $k=\ell$, so that we can rewrite $(1)$ as
$$n=a_km^k+\ldots+a_1m+a_0=b_km^k+\ldots+b_1m+b_0\,.$$
We may assume that $a_k\ge b_k$. Then
$$(a_k-b_k)m^k+a_{k-1}b^{k-1}+\ldots+a_0=b_{k-1}m^{k-1}+\ldots+b_0\,.$$
- This is impossible if $a_k-b_k>0$; why?
Thus, $a_k=b_k$, and
$$n-a_km^k=a_{k-1}b^{k-1}+\ldots+a_0=b_{k-1}m^{k-1}+\ldots+b_0\,.$$
Finally, $n-a_km^k<n$, so $n-a_km^k$ has a unique $m$-adic representation, so $a_i=b_i$ for $i=0,\ldots,k-1$, and the two representations of $n$ in $(1)$ are therefore identical, contrary to assumption. This contradiction shows that no such $n$ exists: every positive integer has a unique $m$-adic representation.