est. San Jose CA · 2026

Yuqiao Simon Su

Software Engineer at Google GCP building AI infrastructure. M.S. Computer Science at Penn. Previously at WeRide (autonomous driving), and earlier at Tencent and Toast.

Currently — SWE L3 at Google GCP
Distributed SystemsAI AgentsLLM Serving
GitHub↗LinkedIn↗Email↗

46 unique visitors

Yuqiao (Simon) Su
Plate I — '26
Experience & Education
work
Google
GoogleSWE L3 · Jul '26work
WeRide
WeRideSWE P3 · Jun–Jul '26work
Tencent
TencentSWE Internwork
Toast
ToastSWE Internedu
UPenn
PennM.S. CS '26edu
Northeastern
NEUB.S. CS '24
Experience
5 entries · click to expand
  • Work2025
    AI Multi-Agent Platform●
    Tencent · Shenzhen, China
  • Work2024
    Wharton Research Data Services
    University of Pennsylvania · Philadelphia, PA
  • School2024
    MS in Computer Science
    University of Pennsylvania
  • Work2022
    Software Engineer Co-op
    Toast · Boston, MA
  • School2020
    BS in Computer Science
    Northeastern University
May 2025 – Aug 2025

AI Multi-Agent Platform

Tencent · Shenzhen, China
Back-end Software Engineer Intern

Built the backend for an AI-driven multi-agent data analysis platform at Tencent, enabling executives to go from natural language questions to SQL queries, visualizations, and business insights — fully automated.

Highlights
  • 01Architected the Coordinator–Worker AI "workforce" architecture to automatically decompose complex analysis tasks, run subtasks in parallel, and aggregate structured reports, charts, SQL, and insights through multi-agent collaboration.
  • 02Developed an Agentic RAG system using Camel-AI OWL for intent recognition and multi-agent coordination; enhanced via NLP tokenization and synonym expansion, boosting recall accuracy by 63%, task success rate by 17%, and reducing LLM token usage by 12%.
  • 03Integrated internal business APIs via MCP Server with unified authentication, fine-grained permission control, and table-level access auditing, expanding LLM access to internal data across company-wide resources.
  • 04Deployed a dynamic knowledge base update mechanism using user feedback and LLM evaluation to continuously enrich a vector DB, enabling the multi-agent system to learn evolving user needs in real time.
Stack
PythonCamel-AIMCPRAGLLMMulti-Agent