Hello there,

I'm Saheb.

I enjoy digging into systems and understanding how they work.

Not just software systems or LLMs, but also computational biology and human psyche.

More recently, I've been focused on AI research, spending most of my time on reinforcement learning and some on mechanistic interpretability.

I'm currently exploring RL generalization, training dynamics, and sampling efficiency.

I also enjoy building and debugging distributed systems; now that also includes transformer architectures, scaling inference, and managing compute infrastructure.