I work on Large Language Models : inference optimization, knowledge editing, and agentic systems.
My current focus areas include:
- LLM Inference : Scalable serving and optimization on modern accelerators
- Knowledge Editing : Post-hoc factual updates to LLMs without retraining (ROME, MEMIT, AlphaEdit)
- Agentic Systems : Multi-agent orchestration, tool-use pipelines, and MCP (Model Context Protocol)
- LLM Evaluation : Dynamic benchmarking frameworks for factual accuracy and temporal reasoning
- Contributor to vLLM : High-throughput LLM serving engine
- Contributor to llm-d : Kubernetes-native LLM inference platform
- Contributor to EasyEdit : Knowledge editing framework for LLMs (ACL 2024)
- Deep Learning-driven Detection of Nuclear Fusion Ignition
- Martian Terrain Classification through Federated Learning
- Exploring Huntington's Disease Diagnosis via AI Models