Currently, the technique to detect neutrinos at the EeV energy scale by the radio signals emitted when they interact in glacial ice, is making the transition from small prototype experiments to full discovery-scale detectors. In the summer of 2021, con- struction started for the Radio Neutrino Observatory Greenland (RNO-G), the first detector large enough to be sensitive to the expected flux of cosmogenic neutrinos. With the prospect of the first detection of a neutrino using the radio technique, the ability to reconstruct their properties from the radio signal becomes more and more important. The goal of this thesis is to develop methods that would allow for the reconstruction of the energy of a neutrino detected by RNO-G or a similar detector. At the same time, radio detection is a well-established technique for cosmic ray observatories. The radio emission from air showers is due to very similar effects, and shares many properties with the radio signals expected from high-energy neutrinos. Thus, air showers can be a useful test case when trying to develop reconstruction techniques for radio neutrino detectors. To reconstruct the energy of an air shower, radio cosmic ray observatories usually use the amplitude of the radio signal measured at a number of different locations. Such an approach is not feasible for a radio neutrino detector, because most events are expected to only be detected by a single radio station. Therefore, the first task was to develop a way to reconstruct the energy of an air shower using only a single detector station. This was accomplished by first using a technique called forward folding, which fits an analytic model to the measured waveforms, to determine the radio signal’s frequency spectrum. By using the shape of this spectrum as a proxy for the angle under which the radio signal is recorded, called the viewing angle, the shower energy can be reconstructed with a resolution of around 15%. The forward folding technique used to reconstruct the radio signal from an air shower relies on an accurate analytic model for the radio signal. This works well for the radio signals from air showers, whose spectrum can be described quite accu- rately by a simple exponential function, but applying it to signals from a neutrino poses some challenges. Their frequency spectrum is more complex and more difficult to predict on a per-event basis. Additionally, because the first radio detection of a neutrino has yet to occur, these models have to rely purely on simulations and are hard to verify experimentally. The solution to this was to use a method called Information Field Theory, and determine the most likely radio signal from the mea- sured waveforms via Bayesian inference. Instead of an analytic model, the spectrum is described by its correlation structure, which makes very few a priori assumptions about the radio signal. The method is in fact shown to be flexible enough to be applied to radio signals from both air and neutrino-induced showers. Finally, these techniques are combined to reconstruct the energy of neutrinos de- tected by a single RNO-G station. The location of the shower is determined using the differences in the signal arrival time between different antennas. The frequency spectrum of the radio signal is determined using the Information Field Theory method and used as a proxy for the viewing angle to estimate the shower energy. The energy of the neutrino that caused the shower is estimated using Bayes’ theorem, taking into account the energy resolution of the detector and the unknown interaction inelasticity.