Keyphrase retrieval, also known as crucial phrase extraction, is a process that extracts crucial terms or phases from texts, facilitating the easy retrieval and analysis of information. This research utilized LinkedIn, which, as of 2020, has vast professional networks and abundant career-related data to extract the relevant vital words most associated with job title extraction from posts related to jobs. For this, two models were tapped: KeyBART from the transformers library and a pre-trained SpaCy Named Entity Recognition (NER). Title extraction from professional summaries is the first application ever performed using KeyBART. In this work, the following approach was used for job title extraction from job descriptions using KeyBART. Titles extracted by the model from the text have been, for the first time, evaluated in terms of quality through the ROUGE score. The results of the fine-tuning of KeyBART inflame such an approach and call for something more calibrated. Next, we created a custom NER model using SpaCy, which was further refined by using a new entity label of “PROFESSION” for custom identification of extracted occupation names with higher accuracy. The training data probably only comes from the title column in the job posting data, so it would closely correspond to how the model works with the data about the job title. We have also tuned parameters for better performance. The proposed model successfully gained 100% in terms of various metrics. Thus, it attests that results are not only highly accurate for extracting and identifying job titles but also great in significance in terms of formant customization of the NER model. The research provides evidence of applicability in customizing NER models across applied domains and methodologies that can be carried to similar challenges in other, more specialized research fields.