The physics CCR group is the Heisenberg group.
I'd follow WP, to fix concepts and notation, as you seem to confuse representations with the Lie algebras they represent.
The three-dimensional Lie algebra $\mathfrak h$ of the Heisenberg group H (over the real numbers) is known as the Heisenberg algebra. (Three generators, so three parameters, a,b,c.)
It may be represented using the space of 3×3 upper-triangular matrices of the form
$$\begin{pmatrix}
0 & a & c\\
0 & 0 & b\\
0 & 0 & 0\\
\end{pmatrix} , $$
with $a, b, c\in\mathbb R $; its exponential is the generic group element,
$$\begin{pmatrix}
1 & a & c+ab/2\\
0 & 1 & b\\
0 & 0 & 1\\
\end{pmatrix} , $$
Note that, in this (defining) representation, the algebra basis elements are not hermitean, and hence the group elements are not unitary. You'd be very wrong if you imagined the group is U(3).
The following three elements form a basis for $\mathfrak h$,
$$
X = \begin{pmatrix}
0 & 1 & 0\\
0 & 0 & 0\\
0 & 0 & 0\\
\end{pmatrix};\quad
Y = \begin{pmatrix}
0 & 0 & 0\\
0 & 0 & 1\\
0 & 0 & 0\\
\end{pmatrix};\quad
Z = \begin{pmatrix}
0 & 0 & 1\\
0 & 0 & 0\\
0 & 0 & 0\\
\end{pmatrix}. $$
The basis elements satisfy the commutation relations,
$$
[X, Y] = Z;\quad [X, Z] = 0;\quad [Y, Z] = 0. $$
In physics, it is also represented by infinite-dimensional hermitean matrices/operators,
$$\left[\hat x, \hat p\right] = i\hbar I;\quad \left[\hat x, \hbar I\right] = 0;\quad \left[\hat p, \hbar I\right] = 0. $$
Note the obligatory physics i upon transition to the hermitean basis! The group elements result from exponentiation with an i and a parameter, resulting into unitary infinite-dimensional matrices.
Further note in this rep that the trace of the central element Z, proportional to the infinite-dimensional identity here, is not zero, even though it amounts to a commutator. This is a standard famously frequent question on the PSE, with a subtle limit resolution.
The unitarity of the representations or not is not a feature of the Group, but, instead a feature of the representation.
Finally, the identity is never a feature of the Lie algebra, but of the universal enveloping algebra. In this representation, it coincides with the center Z.
Best Answer
This is but the π/2 rotation in phase space, a canonical transformation generated by the quantum harmonic oscillator Hamiltonian $H=(X^2+P^2)/2\hbar$. I am skipping the superfluous subscripts j. The structure actually originates in classical mechanics.
That is to say, given $$ [H,P]= i X, \qquad [H,X]= -iP ~. $$
It then follows that, from the Hadamard Lemma (adjoint action), $$ e^{-i\pi H/2}P e^{i\pi H/2}=P-i{\pi\over 2} [H,P] -{\pi^2\over 2!~~2^2 }[H,[H,P]]+...= X,\\ e^{-i\pi H/2}X e^{i\pi H/2}=X-i{\pi\over 2} [H,X] -{\pi^2\over 2!~~2^2 }[H,[H,X]]+...= -P, $$ a right-angle rotation. Intercalating this similarity transformation, further transforms all functions $f(P)\mapsto f(X)$, and $g(X)\mapsto g(-P)$, whence your exponential desiderata relations!
This was first noticed in print by Condon 1937, and serves as the formal underpinning of the celebrated Fractional Fourier Transform.