A stand-alone emulator of ECMWF's land surface scheme (ecLand) has been developed. This emulator, aiLand, uses a multi-layer perceptron architecture, chosen for its balance of accuracy and efficiency and for its differentiability, which is crucial for integration into data assimilation and parameter estimation systems. In this study, we introduce a two-stage learning framework that leverages both synthetic land surface model simulations and real-world observations. We first pretrain the surrogate on extensive ecLand outputs to capture the core dynamical behaviour of key land surface states, evaluating its accuracy, long-term stability, and transferability across variables, depths, and climates. We then fine-tune the pretrained model on in situ eddy-covariance flux observations for selected diagnostic variables, validating against independent flux-tower sites. The pretrained emulator reproduces ecLand's prognostic soil state with a 90-day RMSE of 1.19 K for surface soil temperature and 0.014 m³m^-3 for surface soil moisture, and remains stable over continuous 4-year autoregressive integrations. A single globally trained model outperforms biome-specialist baselines in cross-biome transfer, with residual errors concentrating in snow-insulated cold biomes where an insulating snowpack decouples the soil from atmospheric forcing. Fine-tuning on FLUXNET observations reduces latent heat flux RMSE by 30 % and sensible heat flux by 20 % at validation sites, while preserving prognostic state integrity and improving the physical consistency of the surface energy budget: per-site Bowen ratio error drops by 42 % and energy balance closure residuals fall from 13.4 W m²m^-2 to 3 W m^-2 and below. These results demonstrate that combining physics-based pretraining with observation-based fine-tuning provides a flexible pathway for building accurate, stable, and differentiable land surface emulators suitable for data assimilation and coupled modelling applications.