Vincent Qiao

Appearance

Vincent Qiao ​

Personal Info ​

CTO & entrepreneur who has taken products from zero to scale — driving a startup from $30M to $500M valuation, leading teams of up to 100 across AI, mobile, and web. Building developer tools and AI applications on the side.

Summary ​

  • CTO + Founder + Big Tech combined background, multiple 0-to-1 products (SitIn.ai, LibLibAI, QQ 9.0, Zhiyinlou IM)
  • Full tech-to-business loop: drove company valuation from $30M to $500M, daily revenue from under $10K to $300K
  • AI Agent full-stack experience: built in-house LLM orchestration handling 10M+ daily requests, distilled 14B expert model in production
  • AI Infra cost optimization: ~200 GPU cluster scheduling, reduced monthly server costs from ¥2.2M to ¥500K
  • Billion-level traffic architecture: QQ 9.0 core infrastructure optimization, Electron desktop serving tens of millions DAU
  • Led teams of up to 100 across algorithm, data, frontend, backend, mobile, and DevOps

Career ​

Presence ​

  • Period: Jan 2025 – Present
  • Role: CTO
  • Team: 50-person team including algorithm, data, frontend, backend, Android, iOS, QA, DevOps, and product
  • Business: Global-first AI-Agentic Marketplace (SitIn.ai) — AI Agents analyze users' social media, automatically identify transaction opportunities, and simulate the user to handle the entire workflow from outreach to booking to payment
  • Achievements:
    • Joined when the servers were crashing daily and team morale was low; led a full server architecture overhaul to support 200x business growth
    • Grew daily revenue from under $10K to $300K, reaching an ARR of $60.3M with 75% MoM growth
    • Drove 2 consecutive funding rounds, increasing company valuation from $30M to $500M
    • Peaked at #6 on the US App Store, 33% 30-day retention, Net Dollar Retention over 100%
  • Tech:
    • Built in-house LLM orchestration system handling 10M+ daily requests, powering real-time AI Agent responses across the platform
    • Cross-network memory model enabling Agents to deeply understand user context across social platforms
    • Distilled 14B expert model (Society Model) to reduce inference cost while maintaining Agent autonomous execution quality
    • Dynamic intent recognition and scoring system to identify high-value transaction leads from massive social interactions
  • Links:

SodaArt ​

  • Period: May 2024 – Dec 2024 (7 months)
  • Role: Founder
  • Team: 10-person team covering R&D, product, and operations
  • Achievements:
    • Expanded perspective from a pure tech viewpoint to a holistic company-wide vision
  • Links:

LibLibAI ​

  • Period: Aug 2023 – May 2024 (9 months, company was 1 year old at departure)
  • Role: Tech Lead
  • Team: 50-person R&D team
  • Achievements:
    • Increased market share by 30%, securing the #1 position in the market
    • Reduced monthly server costs from ¥2.2M to ¥500K
  • Tech:
    • AI Image Generation
      • AI cluster management with ~200 GPU scheduling and optimization experience
      • AI acceleration: doubled image generation speed, reduced GPU count from 200 to 110
      • Deep expertise in Stable Diffusion related technologies and optimization
    • International deployment
      • AWS architecture, deployment, and optimization for overseas markets
      • Tackled key technical challenges to reuse domestic GPUs, saving ~¥1M in overseas GPU costs
    • Developed a Figma-like online professional design tool integrated with AI image generation capabilities
  • Links:

Tencent ​

  • Period: Dec 2020 – Jul 2023 (~3 years)
  • Role: QQ Open Platform Tech Lead
  • Team: 30-person R&D team
  • Achievements:
    • Participated in the full iteration cycle of QQ 9.0
      • Department was part of QQ core infrastructure; gained expertise in optimization techniques for large-scale applications
    • Appointed Team Lead within 6 months of joining, then granted Deputy Director authority 6 months later
  • Tech:
    • Backend architecture handling billions of requests, throughput optimization
    • Mobile Open Platform SDK: reduced size from 100MB to 20MB
    • Mobile Open Platform SDK: reduced crash rate from 0.3% to 0.01%
    • Desktop Mini Programs & Mini Games: built from scratch on Electron, serving tens of millions of DAU
  • Links:

TAL Education (好未来) ​

  • Period: May 2016 – Dec 2020 (4.5 years)
  • Role: Zhiyinlou IM Tech Lead
  • Team: 80-person R&D team including frontend, Android, iOS, backend, DevOps, and QA
  • Achievements:
    • Transitioned from individual contributor to engineering team leader
    • Led the 0-to-1 development of Zhiyinlou IM
      • Originally planned for 2-year launch, delivered nearly 1 year ahead of schedule, saving the company ~¥5M
      • Built and motivated the R&D team from scratch
      • Promoted from Level 3-1 to 3-3 in 4.5 years
  • Tech:
    • End-to-end experience building a Feishu-like IM from scratch
      • Familiar with technical architecture of all modules: 1-on-1 chat, group chat, red packets, video conferencing, online documents, calendar management, open platform
    • Proficient in frontend technologies: Node.js, Electron, with performance tuning expertise
    • Proficient in mobile development: Android, iOS, crash fixing, ANR resolution, logging optimization, APK size optimization
    • Proficient in backend high-concurrency optimization: monitoring systems, MySQL sharding, slow query optimization, caching strategies
    • Proficient in DevOps: Docker, Kubernetes, etc.
  • Links:

Early Career ​

  • Niuguwang (May 2015 – May 2016) Frontend Tech Lead. Sole frontend developer handling all company frontend projects; after departure, the company needed an 8-person team to take over.
  • FLTRP (Jul 2012 – May 2015) Senior Java Developer. Promoted to one of the youngest team leads, awarded Best Newcomer twice.

Personal Projects ​

All developed in spare time using Claude Code, and are live in production.

  • vicvic.im: AI-native messaging that treats LLMs and agents as contacts — add them as friends, pull them into groups, and @ them. Its signature "AI Group" lets you @ multiple models in one group for side-by-side answers. Cross-platform Web / macOS / iOS
  • webcc.dev: A web-based terminal interface for Claude Code
  • viho.fun: Lightweight AI CLI tool supporting OpenAI, Claude, Gemini, DeepSeek, Kimi, Doubao and more. Standalone binary — no Node.js required
  • remotion.cool: AI-powered Remotion video creation tool — multiple aspect ratios (16:9 / 9:16 / 1:1), up to 4K resolution, up to 120fps, live preview in browser, one-click MP4 export
  • mcp-servers.online: MCP server aggregation platform for discovering and integrating MCP servers
  • aibaiban.com: AI-powered collaborative whiteboard with Excalidraw and natural language diagram generation
  • ccbot.dev: An AI coding bot that connects Claude Code to Feishu — chat and code directly in group chats, with session management, Markdown card rendering, and one-click PM2 deployment
  • aitubiao.online: AI-powered chart generator supporting flowcharts, architecture diagrams, mind maps, and more

Tech Summary ​