zanwenfu.com  ·  LinkedIn  ·  X  ·  zanwen.fu@duke.edu

Agents are easy to demo and hard to depend on. I build the ones people depend on.

I founded VYNN AI and built every part of it myself. Before that I was an early employee at AutoCodeRover, one of the first coding agents, which Sonar acquired. I've shipped agentic AI at Robinhood and reliability infrastructure for Binance's Web3 Wallet. What I took from all of it: the model is rarely what decides whether an agent holds up. Everything around it is. That's what I build now.

Tellwhen: AI alerts for anything on the web
_{Creator · 2026 · live · how it works}

Tell it what to watch in plain words, and it'll email you the moment it happens. It's built and live; now I'm running its evaluations.

Agent OS: an operating system for AI agents
_{Creator · 2026 · how it works · the thesis}

Every team building agents rebuilds the same plumbing: memory, rollback, checks, budgets. Agent OS is the layer underneath them. A central brain plans the work, each piece runs in its own process, and a monitor that can't touch anything signs off. Every step is a commit in git, so nothing unverified ships and nothing is ever lost.

Robinhood: Agentic AI team
_{Machine Learning Engineer Intern · 2026 · Robinhood Cortex}

I solo-designed and shipped a proactive agent for Robinhood Cortex that decides when market news deserves a customer's attention, instead of waiting to be asked. It cut false positives five-fold in backtesting with no material event missed and scaled coverage 30× at flat latency. I also caught and fixed a production delivery failure that monitoring had missed.

VYNN AI: a personal financial analyst for every investor
_{Founder and sole engineer · 2025 · how it's built · against Claude · agent code}

Ask about any company, fund, coin or prediction market and get a sourced report and a live Excel model in about two minutes, for about three cents. More than 5,000 investors have signed up, and the first 500 came without a dollar of marketing. The model never writes a number. Code computes every figure, and when VYNN disagrees with Wall Street, it stands by its number and shows you both.

AutoCodeRover: one of the first coding agents, acquired by Sonar
_{Early employee · 2024 to 2025 · advised by Prof. Abhik Roychoudhury · IDE plugin · the story}

In 2024, before Claude Code or Codex, it fixed real GitHub issues on its own. I worked on lifting it to 51.6% on SWE-bench Verified and built Self-Fix, which, when no patch will apply, finds the agent at fault and reruns from there. I also built the JetBrains plugin end to end, which merges the agent's fix into the developer's latest code. After Sonar acquired it in 2025, the former AutoCodeRover team's Foundation Agent reached #1 on SWE-bench's unfiltered leaderboard.

At Duke, with Prof. Bhuwan Dhingra, my research tests coding agents, and the AI that judges their work, against real developers. Errata-Bench and Simulator Review, a study of AI reviewers, are coming soon.

• LUMINA · first author. Four agents that screen studies for medical systematic reviews: 98.2% sensitivity across 15 published reviews, with 35× fewer missed studies than a published baseline, at under a cent per citation.
• architectural-damping · in progress. Can poisoned news change what an AI financial agent tells its user? Instruction-style text in 3 of 4 news articles changed the rating in 19 of 100 runs, against 2 of 100 clean; with 1 of 4 poisoned, no more often than clean. Early results.
• football-llm-scaling · in progress. A fine-tuned Llama beats prompting at predicting World Cup results. Controls show the lead is memory of real 2022 scores: fake scores, unseen 2026 matches and hidden team names each remove it. Early results.

• VYNN AI Against Claude: Two Races on NVIDIA: the same words to both, a model in 38 seconds against 7 min 24 s, and what was inside each.
• An Operating System for AI Agents, with Git as Its Memory: git as an agent's memory, and why everyone stops at the OS metaphor.
• Taking AutoCodeRover from SWE-bench to the IDE: putting a research agent inside the IDE without breaking the developer's flow.

_{Teaching: TA for Prof. Robert Duvall at Duke, in three software engineering courses.
Off the keyboard: fifteen years of clarinet.
Making something people want? I'd like to hear about it: zanwen.fu@duke.edu}