Code mixing, the practice of blending elements from multiple languages within communication, is prevalent across various contexts and regions. The motivation for developing a hybrid stemmer for code-mixed Gujarati-English texts arises from the increasing prevalence of code mixing in digital communication, particularly in the fields of social media, educational forums, community discussions, customer service, online marketing, etc. This occurrence necessitates the development of analytical tools capable of proficiently processing and interpreting such hybrid language data to augment comprehension and interaction. This paper presents a comprehensive overview of proposed dataset, which includes a diverse range of code-mixed samples, and detail the architecture of our hybrid stemming model that combines both rule-based techniques and POS tagging to achieve superior performance in text processing tasks. Rule-based stemming involves applying predefined linguistic rules to strip suffixes and prefixes from words to find their root forms. Part-of-speech (POS) tagging is crucial for understanding the syntactic role of words in a sentence, which aids in accurate stemming. For code-mixed text, POS tagging must accommodate the syntactic rules of multiple languages. The integration of POS tagging and rule-based methods provides a promising direction for improving the accuracy and efficiency of stemming in code-mixed contexts. The conclusion suggests avenues for further investigation, such as the integration of deep learning techniques and the exploration of unsupervised learning methods to refine the approach and expand its applicability across diverse languages. Furthermore, incorporating contextual embeddings can significantly improve the performance of stemming algorithms by capturing semantic relationships between words across different languages.