In the last half a century, Computational Fluid Dynamics (CFD) has been established as an important complementary part and some times a significant alternative to Experimental and Theoretical Fluid Dynamics. Development of efficient computational algorithms for digital simulation of fluid flows has been an ongoing research effort in CFD. An accurate numerical simulation of compressible Euler equations, which are the gov-erning equations of high speed flows, is important in many engineering applications like designing of aerospace vehicles and their components. Due to nonlinear nature of governing equations, such flows admit solutions involving discontinuities like shock waves and contact discontinuities. Hence, it is nontrivial to capture all these essential features of the flows numerically. There are various numerical methods available in the literature, the popular ones among them being the Finite Volume Method (FVM), Finite Difference Method (FDM), Finite Element Method (FEM) and Spectral method. Kinetic theory based algorithms for solving Euler equations are quite popular in finite volume framework due to their ability to connect Boltzmann equation with Euler equations. In kinetic framework, instead of dealing directly with nonlinear partial differential equations one needs to deal with a simple linear partial differential equation. Recently, FEM has emerged as a significant alternative to FVM because it can handle complex geometries with ease and unlike in FVM, achieving higher order accuracy is easier. High speed flows governed by compressible Euler equations are hyperbolic partial differential equations which are characterized by preferred directions for information propagation. Such flows can not be solved using traditional FEM methods and hence, stabilized methods are typically introduced. Various stabilized finite element methods are available in the literature like Streamlined-Upwind Petrov Galerkin (SUPG) method, Galerkin-Least Squares (GLS) method, Taylor-Galerkin method, Characteristic Galerkin method and Discontinuous Galerkin Method. In this thesis a novel stabilized finite element method called as Kinetic Streamlined-Upwind Petrov Galerkin (KSUPG) method is formulated. Both explicit and implicit versions of KSUPG scheme are presented. Spectral stability analysis is done for explicit KSUPG scheme to obtain the stable time step. The advantage of proposed scheme is, unlike in SUPG scheme, diffusion vectors are obtained directly from weak KSUPG formulation. The expression for intrinsic time scale is directly obtained in KSUPG framework. The accuracy and robustness of the proposed scheme is demonstrated by solving various test cases for hyperbolic partial differential equations like Euler equations and inviscid Burgers equation. In the KSUPG scheme, diffusion terms involve computationally expensive error and exponential functions. To decrease the computational cost, two variants of KSUPG scheme, namely, Peculiar Velocity based KSUPG (PV-KSUPG) scheme and Circular distribution based KSUPG (C-KSUPG) scheme are formulated. The PV-KSUPG scheme is based on peculiar velocity based splitting which, upon taking moments, recovers a convection-pressure splitting type algorithm at the macroscopic level. Both explicit and implicit versions of PV-KSUPG scheme are presented. Unlike KSUPG and PV-KUPG schemes where Maxwellian distribution function is used, the C-KUSPG scheme uses a simpler circular distribution function instead of a Maxwellian distribution function. Apart from being computationally less expensive it is less diffusive than KSUPG scheme.