About

I’m Rakshith Vasudev, an MLOps/AI infrastructure engineer at Writer, previously at Dell. I build the infrastructure behind modern AI systems, but I’m equally interested in what happens inside the model itself: how architectural choices, training dynamics, data, and evaluation shape what a model can do. My work spans the path from large-scale training and GPU clusters to efficient inference, with a focus on the points where model behavior and system design meet.

This blog is where I take a step back and work through how these systems and models actually work, beyond simply knowing which knobs to turn. I use it to test my understanding, question assumptions I may have picked up along the way, and document the experiments and explanations that make difficult systems easier to reason about.

I don’t see any of this as the final word. I’m aiming to look under the hood, be clear about uncertainty, and write things that are useful to engineers working through similar problems. If you spot something wrong, incomplete, or inconsistent with your experience, I’d appreciate hearing from you.

The best way to reach me is GitHub.