Anurag

I build software that removes friction between people and computers.

Computer Science student, software engineer, and builder of AI-powered tools, automation systems, and desktop applications. Currently studying at GCET, and figuring out what software should feel like.

View Work·GitHub·Resume
Current Status

Most Recent Experience: Development intern at GACL, building industrial OCR systems. Studying Computer Science at GCET. Working on ALEX.

Featured Work

All projects

ALEX

The framework that turns natural language into validated JSON intent routed through a controlled execution layer.

TypeScriptLLMsJSON Schema

GACL OCR Pipeline

Industrial computer vision system built entirely offline. Extracts structured data from dashboard imagery using OpenCV, PaddleOCR, and custom table reconstruction.

PythonOpenCVPaddleOCR

Nexlyra

An Electron-based keyboard-centric browser with an omnibox intent-routing engine.

ElectronNode.jsJavaScript

MinuteMind

Converts meeting discussions into actionable tasks and synced issues. Built at a hackathon.

Next.jsWhisper APIGPT-4o
Philosophy

"Software should adapt to people."

Most software forces people to learn its logic. The interesting problem is the opposite — building systems that understand what people actually mean and handle the rest invisibly. That's what I keep trying to build.

Recent Writing

Archive

Why Most Productivity Software Creates More Work

Most tools optimize for features and organizational overhead, adding friction rather than removing it.

Building AI Assistants on a Laptop That Could Barely Run Them

What constraints taught me about model architecture, lightweight NLU, and optimization during the SAM era.

Losing a Hackathon Taught Me More Than Winning

Why engineering elegance alone isn't enough, and why understanding user context is what actually builds value.