Many scientific and engineering problems involve observing a common phenomenon through multiple heterogeneous sensors or measurement modalities. Such observations typically contain both information shared across sensors, reflecting the underlying system, and sensor-specific or extraneous components arising from measurement processes or environmental effects. Disentangling these contributions is essential when clean or isolated observations are unavailable. We propose a neural autoencoder framework that explicitly separates shared and sensor-specific latent variables from multi-sensor data. The architecture enforces geometric independence between latent components through structural constraints and orthogonality-based regularization, yielding interpretable and disentangled representations. Building on this representation, we introduce a latent-space generative methodology in which generative models are tuned/“restricted” on selected latent subspaces and combined with observed latent variables to synthesize new samples via the trained decoders. This enables controlled data generation with prescribed shared or sensor-specific characteristics and supports cross-sensor inference by producing distributions over plausible measurements in unobserved modalities. We demonstrate the approach on several computational examples, showing effective disentanglement, targeted data generation, and modality imputation in heterogeneous sensing settings.