Senior software engineer, 13 years in. I work across the full arc: greenfield builds, deep legacy modernization, financial-systems domain modeling, performance engineering - where the hard part is usually judgment, not syntax.
- Domain modeling for systems that handle money. Double-entry accounting engines, property-management platforms, billing invariants, domains where getting a model wrong costs real money, and where the second-order consequence of a "simple" feature is usually the whole problem.
- Making old systems safe to change. Strangler Fig, characterization testing, ports & adapters. Turning untested, decade-plus code into something you can refactor without fear.
- AI-augmented engineering. Claude Code as a daily driver, not autocomplete, but surfacing unknown-unknowns across large legacy diffs, scaffolding characterization tests, and spec-driven development with a real harness behind it.
- Spec-driven development that's actually disciplined: requirements traceability, characterization tests, and a harness that keeps AI-generated code honest.
- The right altitude for human judgment when the machine writes most of the code.
- Why "AI will write all the software" takes underestimate the integration and correctness problems that dominate real systems.
- 📍 Chennai, India
- ✉️ [email protected]