AI Engineer · Product Builder · Open to new opportunities

I build useful software where AI meets messy reality.

I'm Divjot Singh Manchanda — an AI engineer who takes ideas from rough prototype to dependable product. I build local-first tools, data systems, and full-stack AI applications, and I'm looking for a team where I can keep doing ambitious, hands-on work.

Let's talk

Explore recent work

Resume

AI applicationsLocal-first toolsData productsSystems that respect users

Selected work

A wide range of problems. The same bias toward shipping.

Civic data / PWA

In active development

Mumbai Rains

A web-first Mumbai rain nowcast PWA that treats live weather information as an evidence problem: first-party routes, radar-history tooling, and privacy-aware crowd reporting.

Cloudflare Workers
TypeScript
Data pipelines

View project

Local-first control plane

In active development

GremlinBoard

A calm, local monitoring board for live widgets, backed by a manifest registry, FastAPI runtime, websocket snapshots, and guarded AI widget generation.

Next.js
FastAPI
AI systems

GitHub

Everyday automation

Blue Star Smart AC Control

A Windows tray app, local web panel, CLI, and optional Android widget for controlling a smart AC without turning a simple task into a cloud dashboard.

Node.js
Windows
Android

GitHub

Privacy-first utility

Samsung Smart Switch Deduplicator

A browser-local cleanup tool that hashes and compares Samsung Smart Switch backups without sending filenames, paths, thumbnails, or file contents to a server.

Vite
File System Access API
Cloudflare

GitHub

Serious systems, unserious subject

Nacho Index

A snack-ranking lab with a weighted scoring model, analytics, review workflows, automated verdicts, and enough rigor to settle a chip debate properly.

Next.js
Cloudflare D1
Product design

View project GitHub

Open slot

A new experiment

There is another idea on the bench. I am keeping the details quiet until it is useful enough to share.

More soon

Coming soon

How I like to work

Curious enough to explore. Practical enough to finish.

I enjoy ambiguity at the beginning and discipline at the end: finding the real problem, building the smallest credible version, then making it reliable. I'm especially excited by AI, developer tools, thoughtful consumer software, and teams that care about craft.

Get in touch

GitHub