I'm a Principal Software Engineer at Advisor360°, India. These days most of my time goes into AI engineering: designing agents, the harnesses they run in, and the guardrails that make them safe to use on real production code.
- Agent harnesses and tooling: Claude Code skills, plugins, hooks and MCP integrations that turn repeatable engineering work into reliable, reviewable agent workflows.
- Multi-agent orchestration: subagent panels that audit, cross-check and verify each other's findings before anything reaches a human reviewer.
- Model routing and governance: running open-weight models through OpenRouter with zero-data-retention guardrails, per-session cost tracking and regional routing controls.
- Agent-driven modernization: using agents to carry a large Nx monorepo through a major Angular upgrade, backed by Playwright verification and evidence-first release sign-off.
Before AI took over my calendar, I spent years building complex, end-to-end web platforms, and that frontend depth is still what keeps my agent work grounded in shipping software.
My current focus is LLM agents, tool use (MCP), evals and agent harnesses, built mostly with TypeScript and Node.js. On the product side I still work daily with Angular, RxJS, Nx, React and Playwright.
