Abstract: The present disclosure relates to general and scalable techniques for modeling in silico the kinetics of systems of connected biochemical reactions. Particularly, aspects of the present disclosure are directed to deconstructing a reaction into a plurality of component steps, translating each component step into a set of rate equations to obtain a standard mathematical construct or model representing each component step, numerically integrating across the standard mathematical constructs or models using a system of ordinary differential equations to determine a contribution of each component step to a rate of change of molecules within reaction, and deriving a in silico behavior of a system utilizing the reaction based on the contribution of each component step to the rate of change of the molecules within the reaction. The standard mathematical constructs or models may be parameterized based on an energy profile for the reaction inferred from machine-learning approaches.