Jaya Krishna Mandivarapu

关于

In my research, I am exploring the latest advancements in Retrieval-Augmented Generation (RAG) and large language models (LLMs) to address key challenges in real-time Computer Vision and Natural Language Processing (NLP) applications. My focus is on developing intelligent agents capable of continual learning, active learning, and cross-domain few-shot learning, spanning both NLP and Computer Vision
domains. This involves enabling agents to learn continually without forgetting, leverage accumulated knowledge effectively, adapt to new tasks with minimal data, and generalize across diverse scenarios. To achieve this, I am also investigating solutions to catastrophic forgetting in neural networks, ensuring that agents retain and build upon prior knowledge over time. Additionally, I am working on a dynamic multi-agent
framework to support a range of use cases, aiming to enhance decision-making capabilities and reduce hallucinations in AI-driven responses, ultimately pushing toward more reliable and adaptable general artificial intelligence.