<p>Nonlinear energy sink (NES) cells are a promising mechanism for suppressing multi-mode vibration of continuous structures; nevertheless, their suppression efficacy is intrinsically governed by the assembly positions and intrinsic parameters. This paper proposes a novel two-step optimization strategy to mitigate this challenge. In the first step, the optimized assembly positions are identified based on locations where NES cells can passively and mostly adsorb the multi-mode vibration energy from the host continuous structures. It inherently treats these locations as sensor placements or master Degrees Of Freedom (DOFs), enabling the reconstruction of the host structure’s multi-mode vibration from these sparse responses. Along this line, the optimized assembly positions are determined by minimizing the condition number of the Modal Assurance Criterion Matrix (MACM), a metric widely adopted to select master DOFs for reduced-order models and to set sensor placements for sparse measurements and reconstruction. In the second step, a time domain response-sensitivity method is employed to optimize NES cells parameters by solving a nonlinear least-squares problem that minimizes the residuals between the calculated and the targeted responses. Numerical simulations demonstrate that modal orthogonality, computed via NES cell positions as master DOFs or sensor placements, serves as a key enabling condition for multi-mode bending and torsional modes vibration suppression. While preserving their capability for multi-mode vibration suppression, NES cells are tuned using the proposed sensitivity-based method to efficiently achieve near-globally optimized parameters, further enhancing their effectiveness in mitigating targeted modes or multi-mode vibrations.</p>