YunhaoJiao

The bottleneck was never writing the code.

Ask most engineers what slows them down and they will point at the blank file, the unwritten function, the feature not yet built. Yunhao Jiao points somewhere stranger: the moment after the code is written, when someone has to prove it actually works. That unglamorous gap - between generating code and trusting it - is the whole business he built. TestSprite, the Seattle company he co-founded in 2023, is an autonomous AI agent that writes test cases, runs them, diagnoses what broke, and proposes the fix. The pitch is almost rude in its simplicity: if AI can write code this fast, why is a human still the slow part of confirming it?

That question is not rhetorical for him. It arrived on a specific night, on an on-call shift at Amazon Web Services, when a small regression slipped past world-class infrastructure and world-class engineers and landed in production anyway. The lesson stuck. Generation had gotten fast. Verification had not. And as AI coding tools began shipping code at a scale no review process was built for, the crack widened into a canyon.

Hangzhou, honors college, and a paper at nineteen.

Before the funding rounds and the customer logos, there was a kid in Hangzhou grinding competitive programming problems. That discipline carried him into Zhejiang University's Chu Kochen Honors College, the kind of selective program that collects the people who treat algorithms like a contact sport. He did research at the University of Michigan, then crossed to Yale for a master's in computer science. Somewhere in that stretch, in 2017, he published his first natural-language-processing paper as lead author - before the master's was even finished.

The detail that says the most about him is not a degree. It is that he co-edited a Chinese high school artificial intelligence textbook. The instinct to make a hard subject legible to people just starting out runs through everything he later built. TestSprite, after all, is a tool that hands the tedious, expert-only work of testing to anyone with an idea and an IDE.

Five years at Amazon, building the thing that checks the things

At AWS he spent close to five years as a senior software engineer. His signature project was the Contract Tests framework for AWS CloudFormation - infrastructure whose entire job is to verify that cloud resources behave as promised across their lifecycle. Read that twice and the company he started later stops looking like a pivot and starts looking like a continuation. He has been building verification systems the whole time. TestSprite is just the version where the verifier writes itself.

Move testing left, all the way to the cursor.

Most teams treat testing as a phase - a downstream gate code passes through on its way to release. Jiao's heresy is to drag it as far upstream as it will go: into the IDE, at the exact moment a line of code is born, inside the developer's own loop. He calls it moving testing left. In practice it means TestSprite plugs into the AI coding environment a developer already uses and validates generated code iteratively, instead of waiting for a human QA pass that can no longer keep up with how fast the AI types.

The early traction backs the thesis. After launching TestSprite 2.0 and its MCP server, the platform grew roughly six times in three months, blowing past 35,000 users on the way to nearly 100,000 developers and QA practitioners. Teams from the largest names in software have shown up - and Jiao, even after raising millions, still posts to the developer community asking for feedback on releases like a founder who has not forgotten who the product is for.

Testing that nobody has to think about.

Ask him where this goes and the answer is almost suspiciously modest: make testing completely hands-off. From generating cases to writing test code to diagnosing failures to proposing fixes - the entire chore, gone, so developers can spend their hours building rather than babysitting. It is an autopilot pitch in a market full of co-pilots. The distinction matters to him. A co-pilot still needs you watching. An autopilot earns the right to be trusted alone.

There is something fitting about an engineer whose first big project verified that cloud infrastructure kept its promises now building a company whose entire reason for existing is to keep AI honest. The code keeps getting faster. Somebody has to stay the conscience. Jiao decided it should be a machine - and that he would be the one to build it.