Sai Mohan Kesapragada

I'm an AI Engineer at Deriv in Dubai. The part of LLM work I care about most is the layer underneath: retrieval that doesn't lie, agents that don't loop forever, and the boring plumbing that turns a model into a product.

Before Deriv I led the build of Arkaa, a production GenAI RAG platform serving enterprise customers. Before that, observability and data infrastructure on the Elastic Stack. The same instinct runs through all of it: I want to know how the system breaks, then build for that.

On the side I run small experiments. Right now: a graph reasoning agent, an Arabic voice mental-health assistant, and an eval harness I keep meaning to open-source. Most of it lives in Python and TypeScript, with Postgres for state and more bash than I'd like to admit.