I am an AI Research Scientist focused on building intelligent systems that are not only powerful, but also reliable, transparent, and useful in real-world settings. My research sits at the intersection of LLMs, multi-agent systems, and trustworthy AI, with recognition at venues including NeurIPS, AAAI, AIED, IROS, and ICRA, including a Best Paper Nomination at AIED 2025.

My current work focuses on accuracy maintenance for long-horizon agentic tasks — designing and evaluating multi-agent harnesses that keep interdependent agents reliable, safe, and coordinated at scale, well beyond where today's frontier models begin to fail. This includes LLM post-training via supervised fine-tuning, RLHF, reinforcement learning with verifiable rewards (RLVR), and evolution strategies with optimized Pass@k reward mechanisms for response diversity and quality.

Across NSF/IES-funded projects and industry deployments, I design and deploy transformer and reasoning models (BERT, GPT, LLaMA), build multi-agent workflows that combine retrieval-augmented generation, knowledge graphs, and tool use via MCP, and evaluate model behavior through human-centered and task-grounded benchmarks. I am particularly interested in bridging fundamental model research with deployable systems that deliver measurable real-world and societal impact — from AI safety evaluation of interdependent agents to AI-augmented assessment and instruction.

My earlier work in distributed robotics and intelligent coordination (including IROS and RA-L publications) continues to inform how I think about collective intelligence, communication efficiency, and decision-making under constraints. This cross-domain perspective helps me connect ideas from robotics, learning sciences, and language intelligence into unified AI solutions.

I actively collaborate across disciplines and welcome partnerships with researchers, labs, and industry teams working on trustworthy LLMs, AI for education, multi-agent intelligence, evaluation, and responsible deployment. If our interests align, I would be excited to connect and co-create impactful research.