✨ Visual Editor

close

palette Canvas & Background

Gradient:arrow_forward
Text Color:
135°

style Card Style

40px
16px

text_fields Typography

16px
kapilansh
@kapilansh_twt
AI lab salaries in 2026:

OpenAI → $640k median total comp

Anthropic → up to $690k for research engineers

Google DeepMind → packages reaching $893k

Meta AI → $100M offers to pull talent from OpenAI

The engineers getting these numbers all have one thing in common

They don't just use AI. They build it from scratch.

"Stanford CS336 – Language Modeling from Scratch 2025" by Percy Liang is the only course that bridges that gap

Free on YouTube. Spring 2025.
Stanford's most cutting edge AI course

You don't just learn how LLMs work. You build one:

• Transformer architecture – built from scratch using only PyTorch primitives, zero shortcuts
• Training at scale – distributed training across GPUs the way OpenAI actually does it
• Flash Attention – the optimization that made modern LLMs possible to train
• Scaling laws – predicting model performance before spending millions on compute
• Fine tuning & alignment – how a raw model becomes something humans can actually use

Everyone is using the black box

The engineers building it are getting $640k+

That's the gap. That's the salary difference
Thread image
kapilansh
@kapilansh_twt
reposts appreciated 🙏🙏🙏
kapilansh
@kapilansh_twt
check this too:

kapilansh
@kapilansh_twt
kapilansh
@kapilansh_twt
kapilansh
@kapilansh_twt
Generated by Thread Navigator
100%
view_carousel Carousel Studio NEW
Press + S to quick-export