Helios
Python · CUDA · Async I/O
An inference server I am building to understand continuous batching, cache management, and scheduling under real constraints.
Software engineer · Mumbai
I build things to understand how they really work.
I spend most of my time around inference runtimes, compilers, and low-level systems. This is where I publish what I learn while building them.
NowTracing where LLM inference spends its time and turning the useful parts into Helios.
Selected writing
All writingSelected work
More detailSysDesignVault
Next.js · TypeScript · D3
A working system-design study environment with simulations and concept maps for the parts diagrams usually leave unexplained.
This website
Next.js · TypeScript · WebGL
A quiet home for my writing, with an interactive terminal available for visitors who would rather explore it from a prompt.
About
I'm a software engineer based in Mumbai. I tend to learn by rebuilding the systems I use: inference servers, compilers, databases, and queues. Writing is how I find the gaps in what I think I understand.