Microsoft
Applied Scientist Intern
Exploring reinforcement learning for agent workflows and translating business-driven research into scalable AI systems.
Experience
Work across research and industry.
Recent work in scientific AI, large language model systems, and production software engineering.
UC San Diego
Graduate Researcher
Researching agentic systems for weather science, including execution environments and tool APIs that let language models write code, run analyses, inspect intermediate outputs, and revise solutions through iterative reasoning.
Built evaluation and synthetic-task pipelines that turn meteorological narratives into executable, verifiable benchmarks for scientific agents.
Scientific AI
Agent workflows
Execution systems
Tencent AI Lab
Machine Learning Engineer Intern
Developed a real-time, multi-stage LLM pipeline for Honor of Kings battle reporting, covering event extraction, narrative planning, grounded generation, and time-aware commentary.
Used GraphRAG, retrieval, and model distillation to improve factual grounding while reducing inference cost under production latency constraints.
GraphRAG
LLM distillation
Real-time serving
Ai4C Applied Research Institute
Software Engineer Intern
Built and maintained a customer-facing platform with a headless backend, GraphQL gateway, and React frontend, deployed through Docker, GitHub Actions, and Kubernetes.
Developed an internal LLM assistant for proposal drafting using prompt templates, lightweight memory, and retrieval over client-specific context.
React
GraphQL
Kubernetes