RE:CZ

Rust+AI: A Summary of Team Backend Migration

Technical Log

👤 Technical team managers and developers focused on performance optimization and technology selection
The team recently migrated the arbitrage project from NodeJS to Rust, resulting in significant performance gains and faster development progress. Midas has also been moved out of Yuan to become a Rust+AI driven project. This stems from the success of the PMA project (200% annualized), proving Rust+AI as a reliable backend. Currently, fund audit code remains on Yuan, with plans to cut it to free up the K8s cluster, but challenges persist. The author increased font size due to eye strain.
  • ✨ Significant performance improvements after migrating the arbitrage project from NodeJS to Rust
  • ✨ Midas project moved out of Yuan, adopting Rust+AI
  • ✨ PMA project achieved 200% annualized return, validating Rust+AI reliability
  • ✨ Residual fund audit code; plans to cut it and free up K8s cluster but difficulties remain
  • ✨ Author increased font size due to eye strain
📅 2026-06-18 · 354 words · ~2 min read
  • Rust
  • Arbitrage Project
  • Migration
  • AI Development
  • Performance Optimization
  • Midas

It's June 18, 2026.

It has been a while since I last wrote a log. Lately, I've been working on strategies for domestic stock index futures.

Recently, I had Ryan migrate the arbitrage project from NodeJS to Rust. The effect was remarkable — it completely solved its performance issues, and now it runs incredibly smoothly. Moreover, since the arbitrage project broke free from Yuan's framework constraints, Ryan's development speed has noticeably accelerated. This proves that smaller and more streamlined projects work better.

Additionally, I also had Midas fully move out of Yuan. Now Midas is also a project driven by Rust + AI development.

I can say that the trend of using Rust within the team has become prevalent, and this all started with the prediction market arbitrage project PMA initiated by Mage. At that time, we held a meeting to discuss the technology choice. None of us knew Rust well (at least not proficiently). We believed that, with the help of AI, writing Rust was achievable. At the same time, spread arbitrage is a project with extremely high latency requirements, so we needed a high-performance option.

After several months of practice, PMA has entered a stage of stable profitability with an annualized return of 200%. Unfortunately, its capacity is too small, and it's unclear how long it can keep running. However, the insight gained remains unchanged: "Rust + AI has become a reliable backend technology choice."

Next, we directly switched the backend of our arbitrage project to Rust within 2–3 days. It now runs on a 2C4G AWS machine, and its performance is excellent — 2C4G is actually more than enough. The entire process was handled by AI development plus AI deployment and operations.

Currently, there are still some fund audit codes lingering on Yuan. I really want to cut them all off and release the entire K8s cluster. However, there are still some difficulties to resolve regarding how to handle fund audits and combined accounts...

Lately, I've been overusing my eyes and often feel visual fatigue. Maybe I'm just getting older... So now I usually set the font size to extra large.

See Also