Synechron · Jun 2021 – Present
Securities monolith, decomposed
Split one application handling securities transactions for investment banks into services, a boundary at a time. Also took the Spring Boot 3 and Java 17 jump, which nobody volunteers for.
- Role
- Tech lead, hands-on
- Team
- [HOW MANY]
- Stack
- Spring Boot · React · Docker · Jenkins
The problem
A monolith handling securities transactions for investment banks had become slow to change and slow to respond. Every release moved the whole application.
What I decided
- Decomposed rather than rewrote. The rewrite was on the table; the transaction logic was the part nobody could afford to get wrong.
- Automated the pipeline before splitting further, so more services did not mean more release burden.
- Took the Spring Boot 3 and Java 17 upgrade head-on instead of parking it behind feature work.
Outcome
25% lower maintenance cost, 20% faster response times, and a clean vulnerability baseline after the upgrade.
How the split actually went
The service boundaries, how the data was divided, what stayed in the monolith and for how long.
What I'd do differently
The paragraph people actually read. Be specific.