An electron, in addition to its electric charge, possesses a small magnetic moment, called spin. The spin of an electron can point parallel (spin-up) or antiparallel (spin-down) to the magnetic field. These two states are analogous to zero and one of the logical bit in current digital electronic devices. However, according to the laws of quantum mechanics, the spin of an electron can be both up and down at the same time. Exploiting the spin degree of freedom has opened up a new era in the field of semiconductor electronics which may revolution current electronic devices. The electron spin could act as a quantum bit (qubit) in a futuristic quantum computer. With recent advances in nanotechnology, it is now feasible to create tiny electrostatic islands called quantum dots to controllably trap single electrons and explore their spin properties. This thesis presents experiments aiming at combining the indispensable ingredients of a quantum computer: read-out, control and initialization of single electron spins. It also seeks a deeper understanding of the properties of single electron spins in GaAs quantum dots. The measurements are performed on a double quantum dot which is defined in a two dimensional electron gas (2DEG) of GaAs/AlGaAs heterostructure. Applying negative voltages to the metallic gates on top of the heterostructure depletes the electron gas beneath them and thereby creates the quantum dots. By applying more and more negative voltages on the gates, we remove the electrons from the quantum dots one by one, reaching the single electron regime. Applying a magnetic field creates an energy difference between the spin states and defines the two states of the qubit. The device used in the experiments is cooled down to about 100\,mK where quantum mechanical behaviour is observed. In the first part of the thesis, we have realized independent single-shot read-out of two electron spins in our double quantum dot. The capability to measure the quantum state of multiple qubits individually and in a single-shot manner is essential for efficient characterization of quantum information protocols. Additionally, in a quantum computer, the result of computation needs to be read-out. The presented read-out method is all-electrical and the cross talk between two measurements is negligible. The read-out fidelities are about 86\% on average. This allows us to directly probe the anticorrelations between two spins prepared in a singlet state, an entangled two-spin state. The independent single-shot read-out of two electron spins also enabled us to fully characterize the operation of the two-qubit exchange gate, an important operation in a spin-based quantum computer, on a complete set of basis states. We observe a deviation of the two qubit gate from the pure exchange which we later account for. In the next step we combine the single-shot read-out with electrical manipulation of single electrons. Manipulation of single electrons can be done using so-called electric dipole spin resonance (EDSR) where the electric field couples to the spin degree of freedom. In quantum dots, EDSR can be mediated in several ways such as spin-orbit interaction, where the spin of an electron is coupled to its momentum, and the hyperfine interaction, where the electron spin is coupled to the nuclear spins of three isotopes of GaAs. We show that at high magnetic fields there is a clearly observable shift in the resonance condition between spin-orbit mediated and hyperfine-mediated EDSR. In these experiments, we introduce adiabatic rapid passage using fast frequency chirps as a robust technique to invert the electron spin in quantum dots. Furthermore, by modeling the EDSR response, we get a deeper understanding of the interplay between spin-orbit and hyperfine mediated driving. These findings could be exploited for enhanced control of dynamic nuclear polarization processes, including selective control of the three nuclear spin species. The focus of the penultimate part of the thesis is to combine single-shot read-out with fast initialization of single electron spins. The capability of fast qubit initialization to a well-known state is crucial for the implementation of the quantum computer for two reasons. First, at the start of the computation qubits need to be initialized. Second, for the error correction schemes, a continuous source of initialized qubits is required where the speed of initialization needs to be faster than the relevant gate operations. In the experiments described in the thesis, we demonstrate electrically controlled fast initialization of a single-spin qubit making use of ``hot spots'' where spin relaxation is enhanced by more than three orders of magnitude. These hot spots occur when the spin splitting matches the quantized orbital level spacing. Voltage pulses applied to the gates defining the double quantum dot allow us to rapidly move to one of the hot spots. There, spin-orbit and hyperfine interactions efficiently mix the spin and orbital excited states and spin-conserving orbital relaxation syphons the entire population to the ground state, thus achieving $\mu$s-scale initialization. In the last part of the thesis, all-electrical independent addressing of a single-electron spins is presented. In those measurements we perform single electron manipulation using EDSR. Surprisingly, we observe well-separated Zeeman splittings in neighbouring quantum dots. This finding provides a direct route to selective addressing of spins in quantum dot arrays without the need for micro-fabricated magnets. The observed splitting also accounts for the deviation of the two-qubit gate from pure exchange, as observed in the first part of the thesis. All results presented in this thesis contribute to meeting the fundamental requirements for physical implementation of a spin-based quantum computer.