The design and optimization of complex chemical processes is a key challenge in chemical engineering and requires knowledge of the underlying kinetic model. This information can be obtained from experiments by inverting the reaction mechanism, which needs to be known therefore. Solving this inverse problem, however, is mathematically challenging, if not impossible, and the reaction mechanism is mostly unknown for novel compounds. Both of these challenges are addressed in the present thesis by proposing a novel methodology for forward reaction model development, which is based on exploring chemical space without the need for prior knowledge. The presently proposed methodology makes use of reactive molecular dynamics simulations to explore the chemistry of gas-phase compounds. In these dynamic simulations the chemical systems are allowed to evolve naturally, based on the atomistic interactions. During this evolution, bond formation and cleavage are traced based on the atomic connectivities and used to detect reaction events. For each reaction, molecular structures are extracted and high-level quantum mechanical calculations are used to predict reliable thermochemistry and kinetics. This novel chemistry exploration scheme is used to generate an ab initio reaction model for the well-studied high-temperature methane oxidation, which is used as a reference. The ab initio reaction model and a novel reaction pathway observed during simulation are validated against this reference and against high-level quantum mechanics. The comparison of the present ab initio reaction model obtained for high temperature methane oxidation to well-established literature reaction model shows striking agreement. This validation case demonstrates the potential of forward reaction model development using the present purely predictive methodology. Moreover, a reaction pathway previously not considered in kinetic modeling is discovered using the present chemistry exploration scheme and successfully validated in a detailed kinetic study. Potential extensions to the presented chemistry exploration scheme are derived, discussed, and implementations are outline. These extensions focus on the inclusion of effects resulting from microscopic balancing: Pressuredependence and reactions involving non-thermal intermediates. Conversion of high-pressure limit reaction models to pressure-dependent models is intended to be described by microcanonical properties obtained via transformation of canonical properties. A similar transformation is used to obtain information about hot reactions, i.e. the kinetics of non-thermal intermediates. Ultimately, these extensions will be implemented in the presently proposed chemistry exploration scheme to obtain even more accurate ab initio reaction models. In conclusion, the present thesis addresses the increasing need for reaction model development of novel chemical compounds by proposing a novel chemistry exploration scheme. The agreement of reaction pathways and rate constants with literature data reveals the potential of trajectory-based chemistry exploration for developing quantitative reaction models. graphs, tables and diagrams