Hi, I'm Yincheng Zhou
CS & Physics @ UPenn VIPER
Building intelligent systems that work reliably at scale.
I am a dual-degree student at the University of Pennsylvania in the VIPER program, studying Computer and Information Science (AI concentration) and Physics and Astronomy. I am passionate about building intelligent systems that work reliably at scale, with deep interests in high-concurrency distributed systems, computer vision, machine learning, and AI agents.
What excites me most is the challenge of making multiple AI agents cooperate to solve problems that no single model can handle, and engineering the infrastructure underneath to stay robust under real-world pressure. Currently, I am putting these ideas into practice as the Founding Software Engineer at Franklink, Inc., where I architect multi-agent pipelines for an AI-native professional network, and as a researcher at UPenn GRASP Laboratory, where I work on sparse depth prediction for event cameras under Prof. Pratik Chaudhari.
I thrive at the intersection of robust engineering and cutting-edge research, and I am always looking for hard problems that live in that space.
Built an iMessage-based networking assistant end-to-end. Designed planner-to-worker multi-agent routing with retrieval + persisted user state using LangGraph.
Developing EventSPD, a sparse depth prediction algorithm for event cameras targeting 2.9x inference speedup over dense baselines for real-time robotics. Advised by Prof. Pratik Chaudhari.
Engineered an NL2SQL service with SQL-first RAG. Exposed FastAPI endpoints that stream SQL, reasoning, and chart-ready results into a production fintech terminal.
Full-stack sentiment platform powered by DeBERTa-ABSA-v2, a model we trained to 82.5% accuracy on per-stock classification from Reddit.
Architected the multi-agent brain behind the first AI-native professional network. Designed async messaging pipeline with high scalability.
SQL-first RAG chatbot with controlled execution. Safely executes SQL queries based on user intent. Smart visualization of results.