Currently, I am an AI/ML Engineer at Country Financial, where I've deployed the first generative AI application in production, serving hundreds of insurance agents.

Research Focus: I'm researching reinforcement learning for training search agents to create plug-and-play solutions for domain-specific RAG applications and Agentic RAG systems. My work focuses on developing lightweight frameworks that optimize search quality without full model fine-tuning.

The key innovation is using "Gain Beyond RAG" as a reward signal - training search agents to retrieve documents that meaningfully improve generation quality compared to naive retrieval. This achieves state-of-the-art performance with 70x less training data (2.4k vs 170k examples) and enables strong domain transfer.

Current Interests:

• Multimodal RAG: Extending search agents to handle diverse data modalities beyond text
• Generative AI Research: Novel architectures and training methods for LLMs
• Reinforcement Learning: RL frameworks for autonomous AI agents and multi-turn reasoning
• ML Classification: Neural networks for document processing and automated categorization

Last updated: January 2025