Managing a million dollar pool with 15k lines of Go

2 months ago

LightSpeed Trading was a crypto trading bot that copied trades from exceptional traders, managing over a million dollars at its peak.
The bot initially used Python but was later rewritten in Go to improve security and performance, doubling trade execution speed.
Despite early success, the project failed due to a crypto market downturn, lack of timely hiring, and insufficient feature development to retain users.
The technical approach prioritized speed over perfection, leading to significant tech debt but allowing rapid iteration and deployment.
Key lessons include the importance of hiring early, focusing on core features that provide immediate user value, and accepting tech debt when necessary for speed.
The final tech stack included a monorepo on an Ubuntu EC2, Postgres, MongoDB, Next.js frontend on Vercel, and Go for backend services.
The project was a valuable learning experience, emphasizing the importance of building something users want and the trade-offs between speed and code quality.

Hasty Briefsbeta