Accurate predictions of population dynamics are critical to the management of natural resources and can be used to inform sustainable harvest policies, to control invasive species, and to guide conservation actions for threatened species. However, accurate predictions of population dynamics require extensive data, which are not always available. Recent statistical methods overcome data shortages by piecing together diverse data types in a single, “integrated” analysis. Integrated models make full use of available data and link different data types directly to the underlying ecological processes. We developed an integrated model to estimate size-dependent survival and fecundity of Murray cod (Maccullochella peelii) from a combination of size-abundance data, mark-recapture data, and individual growth trajectories. We used Hamiltonian Monte Carlo and the TensorFlow software library to generate fully Bayesian parameter estimates, typically assumed to be computationally prohibitive in large integrated models. The integrated model fitted observed size-abundance data closely and generated plausible estimates of vital rates, including fecundity estimates very similar to existing estimates in the literature. Our integrated modelling approach can be used to predict future size-abundance distributions and can be extended to include information on local environmental conditions, individual movement, and interspecific interactions.