Inference for Stochastic Differential Equations has seen a lot of improvement in the recent past with modern hardware and better algorithms. In this talk, we focus on developing a computationally scalable approach to stochastic differential equations using Particle Filters. Particle filters allow us to obtain an estimate of the log-likelihood and using tools from automatic differentiation, we are able to obtain partial derivatives of the log-likelihood with respect to the parameters. We then apply a stochastic optimization algorithm to the parameters to converge to a local optima since most of the parameter trajectories are non-convex. We implement this in a modern high-performance computing framework, JAX which utilizes automatic differentiation and Just-in Time compilation which provides tremendous speed-ups. Finally, we evaluate the performance of this approach on multiple problems and evaluate the results.