We provide a general framework to compute the probability distribution $F_r(t)$ of the first detection 
time of a `state of interest' in a closed quantum system subjected to random projective measurements. In 
our `quantum resetting' protocol, resetting of a state is not implemented by an additional classical 
stochastic move, but rather by the random projective measurement. We then apply this general framework to 
Poissonian measurement protocol with a constant rate $r$ and demonstrate that exact results for $F_r(t)$ 
can be obtained for a generic two level system. Interestingly, the result depends crucially on the 
detection schemes involved and we have studied two complementary schemes, where the state of interest 
either coincides or differs from the initial state. We show that $F_r(t)$ at short times vanishes 
universally as $F_r(t)\sim t^2$ as $t\to 0$ in the first scheme, while it approaches a constant as $t\to 
0$ in the second scheme. The mean first detection time, as a function of the measurement rate $r$, also 
shows rather different behaviors in the two schemes. In the former, the mean detection time is a 
non-monotonic function of $r$ with a single minimum at an optimal value $r^*$, while in the later, it is a 
monotonically decreasing function of $r$, signalling the absence of a finite optimal value. These general 
predictions for arbitrary two level systems are then verified via explicit computation in the 
Jaynes--Cummings model of light-matter interaction. We also generalise our results to non-Poissonian measurement protocols 
with a renewal structure where the intervals between successive independent measurements are distributed 
via a general distribution $p(\tau)$ and show that the short time behavior of 
$F_r(t)\sim p(0)\, t^2$ is universal as long as $p(0)\ne 0$. This universal $t^2$ law emerges from purely quantum dynamics that
dominates at early times.