Information geometry is a combination of classical statistical inference and differential geometry, the idea being that Fisher information defines a distinguishability metric on a statistical model, understood as a parametric manifold of probability distributions. With sufficiently weakly dependent samples, Fisher information per sample approaches the optimal Cramer-Rao bound as the sample size increases. Similar construction exists in quantum statistics, with Quantum Fisher Information defining a metric on the manifold of suitably structured density operators on a Hilbert space whose dimension increases with sample size. This mathematical concept becomes interesting and physically relevant when the density operators are states of an environment or a probe used to indirectly infer parameters of some underlying dynamical quantum process with sufficiently short-term interactions. We first discuss the geometric concepts in a general level, and then focus on the inference of quantum Markov processes through trajectories from an associated continual measurement.