Tailored Resume Generation Using RAG with LLM as per Job Specifications

The traditional resume being common for each specific job applications which result to rejection for job seekers. These research tailors the resume based on different domain of job applications using Retrieved Augmented Generation (RAG) and LLMs. The automated system starts from resume parsing for information retrieval as per the sections using LLM model like Mixtral, Google Gemma and LLAMA-3. The resume has been further stored into vector databases such as Pinecone and MongoDB with JSON resume further retrieved using re-ranking methods such as BM25 and Cross encoder. The prompt techniques used to construct the model to make them understand better. For providing the memory to the LLM models and refining the tailored resume, the conversational buffer memory is implemented. The efficiency and the performance of the methodology is showcase through the performance evaluation metrics such as BERTscore, RAGAs Metrics and Custom metrics such as Content preservation and Job Alignment. By comparing the cosine similarity for content preservation between LLM and RAG with LLM are 72% and 89% respectively. Compared to LLM models using RAG with LLM models generates consistent and relevant tailored resumes. For a better user friendly and seamless UI, the chatbot is developed.