About

I'm Pradhumn Gupta, an ai / llm engineer. I build the unglamorous engineering that makes AI reliable outside of a demo — production RAG pipelines, agentic search, vector databases, and the systems that hold them together when real users show up.

Right now I'm building an AI-powered food discovery platform, where I work on agentic search and retrieval over a large, messy real-world dataset every day. It's where most of what I write about here comes from — production, not tutorials.

I'm a full-stack engineer at heart — comfortable taking something from an empty repo to a shipped product across backend, web, native, ML, and 3D. If it can be built, I'm interested in building it.

Things I've built

From on-device AI to RAG systems and full-stack products — see the projects page.

What I work with

AI / ML: RAG and vector databases (Qdrant), embeddings and retrieval, agentic LLM systems, and on-device ML with MLX and Core ML.

Full-stack web:Python & FastAPI on the backend, React and Next.js on the frontend, plus interactive 3D / WebGLexperiences with Three.js (I completed Bruno Simon's Three.js Journey end to end).

Native: Swift and SwiftUI for macOS and iOS apps.

This site

I write here about what I learn building this stuff: the tradeoffs, the failures, and the decisions that don't make it into tutorials. I try to only publish when I actually have something useful to say.

Work with me

I take on a small number of consulting projects around RAG, search, and LLM systems. The best way to reach me is email or LinkedIn.