Full-stack Developer · Researcher · Music Producer
I build things across tooling, machine learning, and creative work. Right now that means systems software like Munchkin, research on parameter-efficient deep learning with TRIADS, and a growing body of projects, notes, and music that document what I am learning in public.
I am interested in the overlap between first-principles thinking, software systems, and scientific research. A lot of my work starts with a question that seems slightly too annoying or too niche to leave alone, then turns into a tool, a paper, or a longer thread of investigation.
That has recently taken the form of multi-account CLI tooling, productivity software, materials-property prediction research, and ongoing notes around AI, compression, and neuroscience. I care less about sounding finished and more about building work that is careful, testable, and worth revisiting.
Building practical developer tooling with strong systems ergonomics.
Researching small-data, parameter-efficient deep learning architectures.
Publishing work and maintaining a public trail of notes, experiments, and source code.
A mix of systems tooling, research software, and product experiments.
A cross-platform multi-account manager for Codex, Gemini, and Qwen CLIs. It stores sessions in an encrypted local vault, supports fast switching, checks session health, tracks Codex quota state, and can route commands through quota-aware load balancing.
A parameter-efficient recursive attention architecture for materials-property prediction in small-data regimes. The repo pairs code, pretrained weights, and benchmark results with a linked Zenodo preprint.
A distraction-blocking Chrome extension that intercepts YouTube videos and redirects you to an AI-generated summary first, so you decide whether the video is worth your attention.
A web-based mock test platform driven by JSON exam definitions, with timed exams, review flow, results breakdowns, analysis views, and configurable test content.
Preprints, papers, and the working notes behind them.
Preprint published on Zenodo on March 24, 2026. TRIADS introduces a recursive, parameter-efficient architecture for materials-property prediction in small-data settings, combining weight-tied reasoning, deep supervision, and physics-informed features.
Rudra Tiwari · DOI: 10.5281/zenodo.19200579
A 2026 paper published on Zenodo exploring the theoretical impossibility of infinite lossless compression and offering a clarifying framework around the idea.
DOI: 10.5281/zenodo.18404002
A public research notebook and knowledge garden for my notes on AI, compression, neuroscience, transformers, and adjacent topics that feed into longer projects and papers.
Original music is part of this portfolio too. It is where I get to work with emotion, pacing, and atmosphere instead of just systems and papers.
The main places to find my work, writing, and public profiles.
For projects, research conversations, or collaborations.
contactme.rtx@gmail.comSource code, experiments, and public project history.
@Rtx09xResearch identity, authorship, and publication records.
0009-0004-8521-7627Professional profile, current work, and public updates.
View Profile