Accurate and consistent response assessment is essential for guiding clinical decisions and optimizing treatment strategies in brain tumor patients. However, current methods for treatment response evaluation rely heavily on manual assessment of Response Assessment in Neuro-Oncology (RANO) criteria, which is time-consuming and prone to inter-observer variability. To address these limitations, we developed a fully automated pipeline combining segmentation and classification models to assess brain tumor response. Initially, Swin UNETR and U-Net models were trained on the BraTS dataset to automatically segment the whole tumor (WT) and enhancing tumor (ET) masks from FLAIR and WT masked T1c MRI sequences, respectively. Following segmentation from the best performing Swin UNETR model, shape-based and first-order radiomic features were extracted from the longitudinal LUMIERE dataset. A classification model utilizing TabM was developed for classifying the tumor treatment response into one of Complete Response (CR), Partial Response (PR), Stable Disease (SD), or Progressive Disease (PD) classes. Median Dice scores of 0.8811 and 0.8754 were obtained for the WT and ET using Swin UNETR models, respectively on the BraTS dataset. Using the extracted radiomic features, the TabM classifier achieved an average five-fold cross-validation balanced accuracy of 0.6415 on the LUMIERE dataset. A balanced accuracy of 0.5118 was obtained on the hidden multi-centric test dataset comprising 1010 cases of 300 patients. These results demonstrate the feasibility of automatically assessing brain tumor treatment response using longitudinal FLAIR and T1c MRI scans.