LOADING STACK.LAYERS
home stack journey manifesto connect
// 002 — the full vertical

The stack I'm
climbing, top to bottom.

From the physics of a transistor to the math of a transformer — the full vertical. These are the layers I study, hack on, and try to deeply understand. No skipped steps.

LAYER_06 — TOP

06The Unknown

The biggest layer. The unsolved questions, the weird papers at 2 AM, the failures that teach the most. I'm reserving deliberate space to be wrong, often and quickly.

humilitycuriosityfirst principlesresearch taste
VOID.MAP
?
unknown.bin
LAYER_05

05Writing & Learning

Documenting everything. Feynman-style notes, build logs, paper breakdowns — because you don't really know it until you explain it. Public by default.

notesblogsthreadsbuild-in-public
SIGNAL.OUT
Δt · 1.5s
LAYER_04

04Systems & Code

The craft of building. Python, C++, Rust, Linux, distributed systems, type-safe APIs. The boring stuff that makes the magic reliable and reproducible.

pythoncpprustlinuxdockergit
main.rs
// eshaant // build log
fn learn(topic: &str) {
  println!("{topic}");
  while !mastered(topic) {
    build(topic);
    break_it(topic);
    document(topic);
  }
}
learn("neural nets");
learn("silicon");
// → ok
LAYER_03

03Embodied AI

Intelligence that moves. Robotics, sim2real, control loops, sensor fusion — where models meet the physical world and theory meets friction.

ros2isaac simmujococontrolslam
WORLD.SIM
x:0.42 y:0.91
LAYER_02

02Neural Networks

The math of learning. Backprop, attention, transformers, diffusion — building intuition from first principles, not just API calls. Reading the papers.

pytorchjaxtransformersattentiondiffusionrlhf
FORWARD.PASS
L:3 · loss:0.41
LAYER_01 — BOTTOM

01Silicon & Hardware

How bits become thought. Transistor physics, CPU/GPU micro-architecture, memory hierarchies, and what actually runs the model. The foundation.

computer archcudafpgaverilogmemory
CHIP.LAYOUT
7nm · 16-core
// continue

Now go deeper —
the journey.

see the journey → back home ↗