How I Ended Up at a Crypto Exchange

In 2018, I graduated from a three-year junior college. My family used their connections to get me into a home-renovation internet company — not to write code, but in a general helper role.

Three years later, in 2021, I was sitting at a desk at a cryptocurrency exchange, responsible for testing on the core matching engine project. What happened in between was actually pretty simple: I changed jobs twice, taught myself some things, and then took an opportunity most people would have hesitated on.

This article lays out how I got to where I am.

1. My First Job: Home-Renovation Internet

I joined the home-renovation internet company in 2018, brought in through my brother-in-law's connections. I'd originally wanted to be a developer, but back then I was introverted, timid, and bad at expressing myself — having any job at all was already more than I'd hoped for.

The only thing I gained that year was confirmation of one fact: I couldn't stay in a place with no room to grow.

2. The Industrial Internet: Two Years of Full-Stack Odd Jobs

In 2019 I quit with nothing lined up. Two months later I joined an industrial internet company. The company was small with few people, so I was forced to do a bit of everything: testing, ops, writing scripts. I taught myself operations, networking, and architecture, bought my own servers to build test environments, and wrote a fair number of small tools to save time.

Those two years were exhausting, but they left me with one thing: the confidence to solve problems on my own. Two years later, that confidence paid off in a big way.

3. An Offer That Made Me Hesitate

In 2021, I got an interview invitation for a test engineer role at a cryptocurrency exchange.

The hesitation was real: the industry sounded like a gray area, and my family would worry if they knew. But the pay was genuinely high, and the work — derivatives trading, matching engines — was hardcore. I did the math: what I'd always wanted wasn't stability, it was growth. This was an opportunity I might not meet once in ten years in any ordinary industry.

So I took it.

4. First Day on the Job

The company had two to three hundred people. My matching engine project team had a dozen or so, mostly developers, with testers, product managers, and ops each assigned a few seats. I'd assumed testing was just clicking around and filing bug reports.

Then I looked at the system architecture.

Nine services strung into one pipeline by message queues: a single order, from the moment a user clicks "Buy," passes through the access gateway, order service, matching engine, clearing service, asset service, market data... touching seven or eight components. Completely different from the business systems I'd tested before — those were forms, workflows, and permissions, logic that belonged to people. This was data structures, machine logic built to process a flood of orders within a second.

Two concepts made the deepest impression on me:

Matching rules. How do buy and sell orders get matched? By price priority, then time priority — whoever has the better price trades first; at the same price, whoever arrived first trades first; multiple orders at the same price get filled proportionally. This isn't human-brain logic; it's an order book, a priority queue.

Liquidation. When the market crashes, longs get force-liquidated; if the forced liquidation can't execute in time, the position goes bankrupt. And the "mark price" that decides all of this isn't the latest traded price — it's a smoothed, calculated price designed to resist manipulation. The first time I heard that someone would deliberately slam in a huge order to "spike" the price and liquidate others, I sat there stunned for a while: so this is how money disappears in this market.

Every night back then I was cramming: contracts, margin, forced liquidation, bankruptcy, funding rates... It felt like starting over in a brand-new field.

5. After the Initial Confusion

But that confusion didn't scare me off — it fired me up. I was getting my hands on a genuinely hardcore system, where every single bug could translate into real money lost.

Later I came to understand that testing at an exchange carries completely different weight than testing an ordinary business system — you're not testing "can this button be clicked," you're testing "will the system survive extreme market conditions, and will users' money stay safe." That became the most important part of my work for the next two years.

In Closing

The company later shut down over funding problems, and I left. But the matching engine, the liquidation scenarios, and the fund-safety test cases I worked on over those two years are the most valuable things in my career — they turned me from someone who "clicked around" into a tester who writes scripts to dig out system-level bugs.

Coming up in this series:

How we stress-tested the matching engine at 110,000 TPS
Writing my own script to uncover a system-level bug where longs and shorts were liquidated at the same time
A bug that almost started a fight between two developers
What a test team at a two-to-three-hundred-person exchange actually looks like