Extreme weather events involving temperature have become an increasing concern in recent years due to climate variability, especially in India, with a rise in the intensity, frequency, and duration of high temperatures. This work conducts a stationary temperature-wave (TW) analysis in Delhi, using daily maximum temperatures recorded at meteorological stations from 1996 to 2017. The work observes an increasing trend in maximum temperatures in Delhi and investigates this through regression analysis with ensemble techniques. Additionally, it integrates multivariate 1-dimensional Convolutional Neural Networks (1-D CNN), Bi-directional Long Short-Term Memory Neural Networks (Bi-LSTM NN), and LSTMs resulting in the Temperature-wave Analysis: A work with Ensemble Regression Prediction Method (Tilt). The Tilt framework efficiently handles large multivariate time-series datasets and adapts to dynamic temperature pattern changes. The multivariate 1-D CNN captures localized features within the input data, while the LSTM and Bi-LSTM NN construct enduring time-dependent relationships among these features. This framework also predicts maximum temperatures for the hottest years in Delhi. These findings are valuable for various sectors, such as health, urban management, and ecology.