I am a researcher interested in building intelligent agents powered by large language models.
My work focuses on enabling LLM-based agents to reason, plan, use tools, and collaborate
effectively in complex environments.
Research
LLM Agents — my primary research focus
I study how to build autonomous agents that can decompose complex tasks, reason over
multi-step plans, invoke external tools, and recover from errors. My interest spans
single-agent architectures (ReAct, Plan-and-Execute, Reflexion) as well as multi-agent
systems where agents collaborate, negotiate, and share knowledge. I am also interested
in agent evaluation — how do we reliably measure an agent's ability to complete
open-ended, long-horizon tasks?
Tool Use & Grounding
Enabling agents to interact with APIs, code interpreters, databases, and the web.
I explore how agents learn to select, compose, and sequence tools, and how they ground
abstract instructions in executable actions.
Reasoning & Planning
Improving the chain-of-thought and tree-of-thought capabilities of LLMs so agents can
navigate large search spaces, backtrack from dead ends, and produce verifiable plans.
I also work on Large Language Models more broadly.
Papers
Coming soon. Stay tuned!
Projects
Paper2Manim
AI-powered tool that converts research papers (arXiv PDF) into Manim animations.
Three-stage pipeline: scene planning → Manim code generation → MP4 rendering,
with automatic error-correction loops.
Contributor