Extracting entities from resumes is a crucial application of information extraction and a vital component of Natural Language Processing (NLP). With the ever-increasing number of job applications, discovering qualified candidates manually has become a challenging task for employers. Additionally, the scarcity of annotated resume datasets poses an obstacle to entity extraction from resumes. To address these challenges, this paper presents a newly annotated dataset comprising 621 resumes, manually labeled with entities such as name, email, skills, designation, degree, graduation year, employment history, and location. To demonstrate the effectiveness of the annotated dataset, we have proposed a resume parser by utilizing the Spacy and RoBERTa model to extract the significant entities. The experiments are conducted on four sets of datasets 200, 300, 470, and 621 resumes and demonstrated accuracies of 51, 57, 62, and 78%. The proposed annotated dataset and methodology have the potential to significantly improve the efficiency of recruitment processes.