Process
Understand → Disassemble → Architect → Rebuild → Test → Deploy
Six stages, no mystery phases. The order matters: rewriting a system you haven't mapped is how projects die.
- 01
Understand
Before anything is changed, the system gets read: code, data, logs, traffic, and whatever documentation survived. Hard problems are usually just undocumented problems.
- 02
Disassemble
The system is taken apart into its real components — clients, services, protocols, schemas — and mapped, including the parts that were never meant to be looked at.
- 03
Architect
A target architecture and migration path, written down. What gets preserved, what gets replaced, what order it happens in, and what the risk is at each step.
- 04
Rebuild
Work lands in visible increments against real data. Backend, protocol, and data layers first, then the interface, so progress is running software rather than mockups.
- 05
Test
Regression harnesses, telemetry, and instrumentation. Edge cases, bad input, failure paths, and synchronization — verified server-side, because client-side checks are decoration.
- 06
Deploy
Shipped and operated: SSL, process management, backups, logs, monitoring, and restart-on-crash. Handover docs and ongoing support available, never mandatory.
Ground rules
How we work
Documentation is a deliverable
What we learn about your system gets written down and handed to you, not kept in one engineer's head.
You own the code
Repositories, database, and server access are handed over. No lock-in, no hostage hosting.
Direct contact with the engineer
You talk to the person writing the code. No account managers in the middle.