1. LLM Multi-Agents and Trustworthy AI
Exploring the intersection of LLM multi-agent systems with trustworthy AI principles. This research focuses on developing collaborative AI systems that maintain safety, fairness, and reliability while enabling complex multi-agent interactions and decision-making processes.
Details coming soon...
2. Model Merging in Restricted Subspaces
Investigating novel approaches to continual learning and model merging within constrained parameter spaces. I will show with brief posts how to develop optimal low cost efficient techniques for model merging while solving in a constrained parameteric space
Details coming soon...
4. Prompt Optimization and Steering Methods in LLM
Creating systematic approaches for optimizing prompts to enhance LLM performance across various tasks. This work combines automated prompt engineering with theoretical understanding of how prompts influence model behavior and output quality. This work explores the effect optimal prompts have on the steerability of the model during inference
Details coming soon...