rohith@prod:~$ ls -la ./work
I don't list every language I've touched. Instead — here's what I've actually built, and the tools I reached for because the problem needed them.
An end-to-end goals, habits & tasks tracker — with a live MCP server that lets AI assistants run your board for you.
problemTo-do apps track tasks in isolation — never tied to the goals and habits that actually drive progress, and never operable by an AI assistant.
builtA full, deployed product: goals → habits → tasks with subtasks and progress tracking, plan tiers, and an MCP server so AI can read and update your board in real time.
Ask hard questions of dense policy documents and get grounded, cited answers.
problemPolicy and legal documents are long, dense, and unsearchable. Reading them to answer one question wastes hours.
builtA working prototype that reads PDF/DOCX/PPTX and answers in context, with authenticated, persistent chat history — a proof of concept for LLM-in-the-loop document tooling.
A terminal chat server handling many clients at once with sockets and threads.
Speeding up matrix operations by giving each cell its own thread.
Chat with AI models fully offline, no internet required.
A published PyPI CLI tool for moving folders — one command, no flags.
A web voting app with a secure ballot flow and live results.