[guided]We must check that the proposed tangent charts are compatible. The overlap of the two tangent chart domains is
\begin{align*}
\pi^{-1}(U)\cap \pi^{-1}(V)=\pi^{-1}(U\cap V),
\end{align*}
so the coordinate domain for the transition map is $\varphi(U\cap V)\times\mathbb{R}^n$.
Define the base transition map
\begin{align*}
F: \varphi(U\cap V) &\to \psi(U\cap V) \\
x &\mapsto \psi(\varphi^{-1}(x)).
\end{align*}
This is smooth because $(U,\varphi)$ and $(V,\psi)$ are charts in the smooth atlas of $M$.
Now take $(x,a)\in \varphi(U\cap V)\times\mathbb{R}^n$. The point $x$ represents the base point
\begin{align*}
p=\varphi^{-1}(x)\in U\cap V,
\end{align*}
and the vector $a\in\mathbb{R}^n$ represents the tangent vector
\begin{align*}
v_p=d(\varphi^{-1})_x(a)\in T_pM.
\end{align*}
Applying the second tangent chart gives
\begin{align*}
(\widetilde{\psi}\circ \widetilde{\varphi}^{-1})(x,a)
&= \widetilde{\psi}(v_p) \\
&= \bigl(\psi(p), d\psi_p(v_p)\bigr) \\
&= \bigl(\psi(\varphi^{-1}(x)), d\psi_{\varphi^{-1}(x)}(d(\varphi^{-1})_x(a))\bigr).
\end{align*}
The chain rule identifies the second component:
\begin{align*}
d\psi_{\varphi^{-1}(x)}\circ d(\varphi^{-1})_x=d(\psi\circ\varphi^{-1})_x=dF_x.
\end{align*}
Therefore
\begin{align*}
(\widetilde{\psi}\circ \widetilde{\varphi}^{-1})(x,a)=\bigl(F(x),dF_x(a)\bigr).
\end{align*}
Under the standard coordinate identification of linear maps $\mathbb{R}^n\to\mathbb{R}^n$ with matrices, $dF_x$ is represented by the Jacobian matrix $JF_x$. Thus the transition map is
\begin{align*}
(x,a)\mapsto \bigl(F(x),JF_xa\bigr).
\end{align*}
Its first $n$ components are the smooth functions $F_i(x)$. Its last $n$ components are
\begin{align*}
(x,a)\mapsto \sum_{j=1}^n \partial_{x_j}F_i(x)\,a_j,
\end{align*}
for $1\le i\le n$. These are smooth because $F$ is smooth, so all partial derivatives $\partial_{x_j}F_i$ are smooth, and multiplication by the coordinate function $a_j$ preserves smoothness. Hence the transition map is smooth.
Now define the reverse base transition map
\begin{align*}
G: \psi(U\cap V) &\to \varphi(U\cap V) \\
y &\mapsto \varphi(\psi^{-1}(y)).
\end{align*}
This map is smooth because it is the transition map from the chart $(V,\psi)$ to the chart $(U,\varphi)$. For $(y,b)\in \psi(U\cap V)\times\mathbb{R}^n$, let $q=\psi^{-1}(y)$ and $u_q=d(\psi^{-1})_y(b)\in T_qM$. The same chain rule, now written out for the reverse direction, gives
\begin{align*}
(\widetilde{\varphi}\circ\widetilde{\psi}^{-1})(y,b)
&= \widetilde{\varphi}(u_q) \\
&= \bigl(\varphi(q),d\varphi_q(u_q)\bigr) \\
&= \bigl(\varphi(\psi^{-1}(y)),d\varphi_{\psi^{-1}(y)}(d(\psi^{-1})_y(b))\bigr) \\
&= \bigl(G(y),dG_y(b)\bigr).
\end{align*}
Under the standard bases, $dG_y$ is represented by the Jacobian matrix $JG_y$, so the reverse transition has formula
\begin{align*}
(y,b)\mapsto \bigl(G(y),JG_yb\bigr).
\end{align*}
Its last $n$ components are
\begin{align*}
(y,b)\mapsto \sum_{j=1}^n \partial_{y_j}G_i(y)\,b_j,
\end{align*}
for $1\le i\le n$, and these functions are smooth because $G$ is smooth. Therefore the inverse transition map is smooth.[/guided]