https://www.linkedin.com/in/suryasure

https://twitter.com/suryasure05

https://github.com/SurjaHead

https://medium.com/@suryasure734

https://decompose.substack.com/

I have a strong interest in understanding the fundamentals of hardware and how we can map ML math to silicon.

I believe the best way to truly understand complex systems is by simply building them, so most of my energy is currently spent on building and understanding different elements of the hardware x ML intersection.

Currently, I’m:

• Studying electrical engineering and getting really good with reverse engineering problems
• Building a minimal TPU to do inference + training on-chip
• Working towards benching 225

Previously, I’ve:

• Built a softmax module in SystolicVerilog
• Built a fully programmable 16-bit computer in hardware description language (HDL) and assembly
• Experimented with introducing generative AI features to over 5 million users at Hevy
• Built a hand-written digits classifier in NumPy
• Programmed an EEG headband to detect focus levels

Check out more of my work below: