Wenzhao Hu

Hi, I’m Wenzhao. I am a Systems Engineer and Architect specializing in high-performance computing and distributed state.

My career began in the browser, where I architected complex synchronization engines and rendering systems for massive-scale productivity suites like Univer and Lark. I have since pivoted to Agent & AI Infrastructure, applying my background in user experience, latency optimization and distributed architectures at NVIDIA.

Experience

NVIDIA — Senior Software Engineer (AI Infrastructure), 2025 - now

Building DGX Cloud Lepton using TypeScript & Golang.

DreamNum — Tech Lead & Architect (Univer), 2023 - 2025

Engineered a real-time synchronization engine based on Operational Transformation (OT) to solve complex distributed state consistency challenges.

  • Architected a plugin-based, isomorphic framework running across browser, Web Worker, and Node.js
  • Engineered a real-time OT-based collaboration engine supporting offline editing and multi-document sync
  • Designed an RPC layer for cross-thread and cross-process communication
  • Prototyped a multi-agent data analysis system using LangGraph, enabling intelligent insights within the spreadsheet engine.

ByteDance — Software Engineer (Lark Sheet), 2021 - 2023

  • Co-designed next-generation architecture for Lark Spreadsheets
  • Improved codebase maintainability and feature delivery velocity

Tencent — Software Engineer (WeCom Document), 2020 - 2021

Optimized initialization sequences to cut document load times by 10x (from >10s to <1s). Implemented high-performance canvas rendering and dependency injection systems to minimize memory overhead.

  • 10x performance improvement on document loading (>10s → <1s)
  • Built high-performance canvas rendering pipeline
  • Implemented dependency injection system to reduce memory footprint

Works

Read more about my works here.

Talks

X  · GitHub  · LinkedIn  · wenzhaoh@nvidia.com