<p>This study presents an integrated artificial neural network (NN)—based predictive modelling and meta-heuristic optimisation framework for multi-response optimisation of wire electro-discharge machining (Wire EDM) parameters during machining of Hastelloy C-276. Due to its high strength and low thermal conductivity, Hastelloy C-276 is difficult to machine using conventional techniques, necessitating advanced optimisation strategies. Six process parameters—pulse-on time, pulse-off time, arc-on time, arc-off time, wire feed, and servo voltage—were investigated using a Taguchi L27 orthogonal array. Four key performance responses were evaluated: material removal rate (MR), kerf width (WK), surface roughness (S), and recast layer thickness (R_LT). An NN model was developed to predict machining responses, and its weights were optimised using four meta-heuristic algorithms: Genetic Algorithm (G-A), Particle Swarm Optimisation (P-S-O), Grey Wolf Optimisation (G-W-O), and Whale Optimisation Algorithm (W-O-A). Model performance was assessed using R², RMSE, and MAPE metrics. Among the hybrid models, NN-G-W-O demonstrated superior predictive accuracy and convergence stability. Comparative analysis with static multi-objective decision-making techniques (M-O-O-R-A and AHP) indicated improved robustness of the hybrid optimisation framework within the investigated parameter space. Sensitivity analysis and Pareto optimisation further clarified parameter influence and trade-offs among responses. The proposed framework provides a reliable and systematic approach for multi-response optimisation in advanced machining applications.</p> Graphical Abstract <p></p>