Bhupendra Bajiya

Developed an advanced GraphRAG system leveraging Neo4j Graph Database, LlamaIndex, and LangChain to deliver highly efficient graph-based data retrieval and generate precise, well-cited responses. Integrated ColPali into Retrieval-Augmented Generation (RAG) systems, enhancing multimodal content retrieval by jointly processing text and images, improving accuracy and relevance in complex document search. Integrated LightRAG, enhancing retrieval accuracy and cost-effectiveness, with integrated Neo4j and MongoDB for efficient data management, and enabling responses with streaming and well-cited information.

Implemented GraphRAG using LlamaIndex with Neo4j, enabling efficient graph-based storage, retrieval, and context-aware answers with precise citations. Leveraged GLiNER and Sentence Transformer for key term extraction and similarity scoring, improving document relevance. Achieved robust sentiment analysis with PyABSA and utilized the T5 model for text paraphrasing and summarization. Integrated Reinforcement Learning for the retail sector, implementing dynamic pricing optimization using Q-Learning and Deep Q-Networks (DQN) for improved decision-making and profitability.

Developed a Mask R-CNN-based system to detect topic shifts in educational videos, enhancing segmentation and navigation, improving user engagement and learning outcomes. Implemented and fine-tuned LLaMA2 for text summarization, generating concise and accessible educational content summaries that improved comprehension and user experience. Integrated Tesseract OCR for text extraction and LaTeX-based OCR for detecting mathematical equations.