I build software and research agentic systems, with a focus on local AI.
I'm a research assistant at ELITE Research Lab LLC, with earlier engineering and QA internships at Genesys.
My interests centre on agentic systems and local AI, including how agents use memory and tools, and how to make them reliable and secure. I'm also interested in GPU inference with CUDA and low-level AI systems in Rust, C and C++.
In my paper on MeReader, I explore how local models can answer questions about a book using only the part you've read.
B.Sc. Computer Science, Eötvös Loránd University, Budapest. Graduated in 2025 with a CGPA of 4.49/5.
I work on Arch Linux with Neovim and a terminal, using Codex, OpenCode and Claude Code for AI-assisted development.
Sep 2026 to present
Jun 2026 to Aug 2026
I study how AI agents use memory across sessions and build reproducible experiments to evaluate their behaviour. I also contribute to shared research resources and tools for working with AI agents.
Worked on customer journey management workflows, from business logic to tests and bug fixes.
Built end-to-end test coverage for Angular and Vue releases.
Guided first-year international students through programming fundamentals and lab work.
An offline reading assistant that keeps retrieval and generation within the reader's current position, so future content stays out of its answers.
The paper evaluates local models on reading comprehension and response time, and examines whether answers stay within the reader's progress.
Mohammed Efaz, Udo Bub, Itilekha Podder
My ORCID profile
https://orcid.org/0009-0002-3562-7607
Built a private reading system that limits retrieval and generation to the reader's current progress, using hybrid retrieval to find relevant passages.
I conceived and built MeReader as the research system for the paper.
A Tauri and React e-book reader with typographic controls, responsive layout, and sync across devices.
I contributed to an existing open-source reader. I didn't create or own the project.
A Vue and TypeScript Chrome extension that summarises Gmail messages, articles, and PDFs locally, with questions about imported content.
I conceived and built MeTLDR as a local summarisation tool for long-form reading and research.
Light-motorcycle game with deterministic movement and collisions.
Home automation simulation with sensor-driven control loops.
Trading card platform with a PHP backend and browser interactions.
Gas-layer simulation implemented with object-oriented design patterns.
Browser map game with terrain placement and seasonal scoring.
I use TypeScript for web applications and Python for AI research. My work also includes automated testing and Linux development.