Stress is the body’s reaction to internal or external pressures, triggering physical, emotional, and behavioral responses, while it can enhance focus temporarily, prolonged or excessive stress can negatively impact mental well-being and overall health. This paper proposes a hybrid deep learning model combined using the strengths of LSTM and 1D-CNN architectures for the stress detection and classification task using HRV data. The 1D-CNN path extracts spatial features and identifies correlations within the HRV metrics, and the LSTM path models the temporal dependencies and captures sequential variations in physiological signals. These approaches integrated into the hybrid model, it reached a accuracy of 88.5% and was capable of classifying stress levels into low, medium, and high. Notable features such as MEAN RR, SDRR, and LF_HF played an important role in capturing both spatial and temporal patterns. The model will be robust and show minimal overfitting along with high generalization, placing it as one of the hopeful solutions for the real-time stress monitoring in workplace and healthcare. This work establishes a basis for advanced stress detection technologies within intelligent healthcare and personalized wellness solutions.