The emergence of the Low-Altitude Economy, encompassing UAM, eVTOLs, and logistics drones, demands a paradigm shift in aerodynamic design. Unlike traditional aviation, this sector requires rapid iteration of thousands of configurations. However, a critical bottleneck remains: high-fidelity Reynolds-averaged Navier-Stokes (RANS) simulations are prohibitively expensive for such large-scale sweeps, while low-cost Euler solvers fail to capture essential viscous phenomena like flow separation and drag. To resolve this tension, we present a data-driven framework that reconstructs high-fidelity RANS fields directly from low-cost Euler simulations. We propose a physics-aware Transformer architecture as Figure 1 utilizing a "slice-tokenization" strategy, which aggregates grid points into physically correlated tokens to reduce attention complexity. This architecture creates a synergistic mapping: Euler solutions provide global physical priors, while explicit positional encodings and condition modulation guide the precise reconstruction of viscous-dominated regions. Evaluated on transonic supercritical airfoils, our method reduces reconstruction error by up to 48% for streamwise velocity and 25% for pressure compared to baselines. Crucially, the model recovers complex viscous physics, such as shock-boundary layer interactions and no-slip wall conditions, from inviscid inputs. By enabling RANS-level accuracy at near-Euler computational costs, this framework offers a scalable "digital wind tunnel" for the low-altitude economy, significantly accelerating the design and certification of next-generation aerial platforms.