About

Premesh Devasani

I'm a data engineer and AI systems builder focused on the intersection of reliable data pipelines, modern LLMs, and real-world analytics. I build things end-to-end — from raw data generation to production-ready dashboards and AI agents.

What I'm focused on

  • Building analytics infrastructure that scales — BigQuery, dbt, and Python pipelines that don't need constant babysitting
  • AI-augmented analytics — using LangGraph and Claude to turn natural language questions into precise SQL answers with charts
  • Shipping production-quality work — not notebooks or prototypes, but deployed systems with auth, error handling, and real data

Data Stack

  • Google BigQuery
  • dbt Core / Cloud
  • Python (Pandas, Faker, SQLAlchemy)
  • Apache Spark (PySpark)

AI & Apps

  • Claude (Anthropic API)
  • LangGraph, Vanna
  • Next.js + TypeScript
  • Vercel AI SDK v6