I am a Computer Science graduate student at The University of Texas at Austin, currently affiliated with the Urban Information Lab, where I work on city-scale AI agents and multimodal retrieval-augmented systems.
My research interests lie at the intersection of large language models (LLMs), multimodal reasoning, and real-time agent systems. I focus on developing context-aware agents that bridge perception and knowledge, especially in domains.
- Large Language Models — inference-time augmentation, tool use, LangGraph workflows
- Multimodal AI — hateful meme classification, vision-language grounding (BLIP, CLIP)
- System Architecture — RAG pipelines, LangChain, MCP (Model Context Protocol), FastAPI
- Applied ML — Reinforcement learning, classifier robustness, AUROC-based evaluation
- SafeMate: A multimodal RAG agent for emergency preparedness using MCP, LangGraph, and streaming LLM interfaces
- Adaptive Traffic Signal Control: Multi-agent RL using SUMO, DQN, and TraCI for traffic flow optimization
- Hateful Meme Classification: Integrating RAG-based knowledge retrieval with image-text classifiers
- LLM-driven Crypto Trading Agent: Strategy modeling and execution via OpenAI and Gemini APIs
- Languages: Python, C++, Dart, JavaScript, SQL
- AI/ML: PyTorch, HuggingFace Transformers, SentenceTransformers, BLIP, CLIP
- Full Stack: FastAPI, LangChain, LangGraph, Flutter, React
- DevOps: Docker, Railway, Firebase, EC2
2025: Submitted: SafeMate: A Modular RAG-Based Agent for Context-Aware Emergency Guidance2025: Submitted: MemeInterpret: Towards An All-in-one Dataset for Meme Understanding
“Context-aware, multimodal AI is the future — especially when grounded in real-world complexity.”