About

I did not begin with a narrow research identity. I spent several years building across web applications, backend services, deployment infrastructure, Linux tools, and applied AI. That breadth was useful: it showed me that the problems I found most interesting were not confined to a particular language or application.

I was repeatedly drawn to what happened when resources became constrained: memory pressure, wasted computation, unstable performance, and the difference between a system that merely works and one that remains usable under stress. That interest led me toward systems research.

My current work studies memory pressure on Linux. I have validated early PSI/swap telemetry and am now evaluating whether sustained cgroup-scoped PSI can safely activate compressed memory earlier under limited RAM. More broadly, I am interested in resource management across operating systems, runtimes, and cloud systems.

My earlier research focused on Nepali legal information retrieval and AI-assisted research workflows. It gave me experience designing evaluations, working in a low-resource language setting, and communicating empirical results. I now bring that experience to systems work through controlled workloads, repeated trials, stable experimental artifacts, and explicit limitations.

I also build systems software. Strata and admux reflect my interest in inspectable state, durable workflows, and tools that remain understandable outside the environment in which they were created.