About

Understanding systems from the inside

I spent nine years in software development in test and quality engineering. As an SDET, I worked with the code and internals of complex applications. I wrote tools and test systems, reviewed code, and investigated failures. I was a test architect for Postman and Acceldata, then head of quality at KnackLabs.

That work gave me a grounding in system design: how components fit together, how applications behave as they scale, and where a design trades performance, reliability or simplicity for another requirement. Whether choosing a language or database, deciding where events help, or weighing a monolith against microservices, I start with the application's needs, leave room for growth and add complexity only when there is a concrete reason for it.

Engineering with AI

I now lead engineering at KnackLabs. My main area of work is enterprise AI systems: deciding where AI is useful, designing the software and operating controls around it, and helping teams deliver it.

My experience with system internals helps me build faster with coding agents. I can break a problem into concrete changes, explain the architectural constraints, and judge whether an implementation fits the wider system. I bring that approach to the engineering teams I work with. In 2025 I documented my early working practices for repository context, acceptance criteria, status, and tests. Symphony Forge developed those ideas into the process our engineers use to build and review software with agents.

Developing the team

I lead engineering training at KnackLabs, create playbooks, and mentor the team. Helping engineers develop their skills is part of my role alongside setting technical direction and delivering projects.

Products and open source

I wrote the initial Firecrawl MCP server and the initial JavaScript LangChain MCP adapters. I also contributed a voice provider, a host-only bash command, and a tool-failure fix to OpenClaw. I built Gantry for operating AI employees with roles, controlled access, approvals, and audit records.

How I work

I start by understanding the work people need to complete and what gets in their way. Our forward-deployed engineers work directly with customers, at their location when needed, to observe the process, uncover undocumented decisions, and inspect existing data and systems. We agree on the problem and how we will judge the result before choosing a solution.

I set the engineering direction and review the important decisions. We start with the smallest change that addresses the problem, validate it against representative work, and refine it from what we learn. Delivery includes clear access boundaries, human approvals where needed, and ownership of the system in production.

Speaking and community

LangChain Ambassador, and a host of the LangChain Connect India meetups in Hyderabad and Bengaluru. I co-host Claude Community meetups in Hyderabad. The first ran from the KnackLabs office, and AI for Leaders was an invite-only roundtable with Anthropic. I was a mentor at the Hyderabad leg of the World's Largest Hermes Buildathon in July 2026. My talks include sessions at The Fifth Elephant, HydPy and LangChain Connect.