← back

Pulsar

SSD-streaming inference engine for giant MoE models (Rust + CUDA). GLM 5.2 743B at 2 tok/s and Hy3 295B at 7 tok/s on two consumer 16GB GPUs. Zero-config multi-GPU: measures PCIe bandwidth, places attention and hot experts where they fit.

https://github.com/giannisanni/pulsar ↗
Thesis fit
Off-thesis

Outside your stated focus; easy to skip unless something exceptional.

Edit thesis
Below your usual filter (40+)
Idea match Light

How close the company’s idea is to your thesis statement

Sector None

Overlap with sectors you care about

Geography Unknown

Location unknown — scores 0

Your thesis: “We back exceptional technical founders building AI-first products and infrastructure, deploying $100K checks within 24 hours.”

Founder ↑ improving Traction ↑ improving Idea vs market ↓ declining
Generate memo
Add / edit details

Correct facts used on the next screening or memo.

Activity & evidence

GitHub giannisanni/pulsar · owns repo · 9.0★/day signal ▸

Similar baseline plays (YC · idea space)

ParaQuery · Active
Managed GPU-accelerated Spark + SQL at 2x speed and half the cost
founders not scraped yet
Expanse · Active
Unlock wasted GPU capacity.
founders not scraped yet
TensorPool · Active
Vercel For GPUs
founders not scraped yet
Thunder Compute · Active
VMware for GPUs
founders not scraped yet
Lilac · Active
We automatically monetize idle GPUs
founders not scraped yet