I will say this again: Mellanox was the most consequential acquisition of this century

Bruh

super juiced to be playing with these. the potential is high

“With the surplus of your labor” is a wild qualifier to add lol. I know it’s Twitter but cmon at least justify what you’re saying a lil bit 😭

Very excited to be working with M13 and Karl! We're off to the races now.

.M13 has been an amazing partner in our seed round. A thoroughly technical team that understands the problems that we’re trying to solve. High performance GPU code needs automation and abstraction. That’s exactly what we’re going after at Mako!

check it out ma

Oh and we raised $8.5M from M13, Neo, Flybridge, Jeff Dean and more Join us to work on the hardest problems in code generation. Instant interviews for those with: - kernel writing experience (CUDA, Triton, HIP) - reinforcement learning (esp large scale) - good vibes

i shouldn't be writing our twitter posts honestly

woohoo

MakoOptimize autotune config flags and env variables for vLLM and SGlang. There are literally billions of possible configs. One of them is the fastest.

Another EZ banger

Mako (Mako) banks $8.5M Seed (M13) to obliterate #GPU tuning bottlenecks. Waleed Atallah, Mohamed Abdelfattah, and Łukasz Dudziak built AI that generates and optimizes kernel code faster than your best engineer ever could, across NVIDIA, AMD, and any #cloud you want. The

New startup idea: a gym that sneaks tren into the water so everyone gets jacked and sees results and retention is 100%

I Am Very Smart™

ludwig “How to scale your model on TT” guide coming soon

Oh it’s ON on. AI at AMD 🤝Mako

I will see you at hotchips hotchipssymposium.regfox.com/hot-chips-2025

MCP server for Claude Code and Cursor integrations Use the agents you love, but let ours write the gpu kernels