// surface level — begin here

where the ocean
meets the kernel

on autonomous machines, the mathematics of wonder,
and what lives at the edge of what we can build.

read the transmissions  →
surface ↓ depth
recent dives
001
How I cut a GStreamer pipeline from 700ms to 120ms
Every millisecond is a meter when your machine is underwater. Here's the architecture decision that changed everything — and why a maxlen=1 deque is not a hack, it's physics.

I'm Arnav — I build ML inference pipelines for autonomous machines and spend the rest of my time trying to understand why the universe works the way it does. This is where those two things meet.

edge inference ROS2 autonomous systems GStreamer neuromorphic computing