Rares P. / Software engineer

Give me the messy system. I’ll find the useful path through it.

I build software for complicated systems: tools people can use, boundaries they can trust, and games worth getting lost in.

Illustrated tired operator holding an orange coffee mug at a computer
operator.state calm under ambiguity
coffee nominal

Focus

Systems, security & applied AI.

Open-source maintenance and independent software.

Approach

Spike. Iterate. Stabilize.

Find the constraint. Build a working path. Make it last.

Tools

Python for glue. C for guts.

Linux underneath. Games on the side.

Recognition / September 2026

A small loop. A first-place finish.

White Approach won first place overall in Micro Jam 064: Mountains! This certificate was awarded to Small Loop Works.

Explore White Approach & the game shelf
Certificate awarded to Small Loop Works for first place Overall Entry in Micro Jam 064: Mountains!, issued September 2026 View the certificate

01 / Selected work

Built, tested, maintained.

Software across systems, security, AI, and native applications. Filter by area, or show the contribution behind each project.

02
Private engineering tool Python / Podman / Git

Review Ops

Private tooling for repeatable code reviews. It pins Git ranges, checks artifacts, and runs probes in offline containers with no secrets.

My contribution: evidence-bound findings, bounded subprocesses, stale/duplicate guards, immutable probe inputs, and human-controlled publication.

trust boundariesreview systems
Private / described at system level
03
Cross-platform desktop PySide6 / audio / AI

DJ Blue AI

A cross-platform desktop assistant for audio playback, transcripts, and Gemini-powered chat, with persistent settings and automated releases.

My contribution: brought UI, playback, transcription, configuration, packaging, documentation, and Linux/macOS/Windows release paths into one operable desktop product.

desktopapplied AIrelease
04
ML platform PyTorch / RunPod / Vast

Chess Bot

A complete chess ML workflow: validate data, train and evaluate models, manage cloud GPUs, and play against them in a standalone arena.

My contribution: connected data generation, training, evaluation, inference, UI, cloud execution, tests, committed behavior specifications, and a separate CPU demo deployment with draggable-board play.

PyTorchpipelinescloud GPUarena
05
Local-first agents LangGraph / Qdrant / RAG

Local Agent Tools

Local-model agents with explicit routing, checkpoints, retrieval, and controlled tools. Built with LangGraph, Qdrant, and llama.cpp.

My contribution: explored agent orchestration as an explicit state machine with local models and controlled tool boundaries rather than a chat wrapper.

agentsRAGlocal LLM
View source
06
Small native product C / SDL2 / CMake

Clean Mines

Minesweeper for desktop and terminal, built around a compact C core with self-tests, portable builds, and automated releases.

My contribution: evidence that the “C for guts” part is real: deliberately small scope, explicit memory/runtime boundaries, and portable builds.

CSDL2portable
View source
07
Playable game Godot / simulation / releases

Code Review Simulator

Review pull requests from a physical apartment. Earn upgrades, automate the work, and live with what the automation gets wrong. A Godot game taken from prototype to playable release.

My contribution: gameplay systems, diegetic UI, performance work, cross-platform builds, release automation, screenshots, public devlogs, and a playable browser build.

01game systemsshipping
Play / view devlog

Small Loop Works / Games

Small loops. Playable worlds.

Mountain climbs, unlikely coffee, and systems with a mind of their own. Browse the shelf and play in your browser.

Explore the games

02 / How I work

Find the constraint. Close the loop.

Start with what is unknown. Test a small working path, then make it dependable. Explore each phase of the process.

question / 01

What is actually unknown?

Map the system and trust boundaries. Reproduce the risky behavior. Build a small probe to test the assumption.

  • Map the system and its boundaries
  • One thin end-to-end path
  • Record what the experiment shows

03 / Review & perspective

Follow the behavior through the system.

A patch is one part of the story. I follow what should happen through code, stored state, permissions, and tests, then reproduce the failure and check the smallest safe fix.

Read the plain-text CV
  1. 01
    Reconstruct

    State the intended invariant and ownership boundary.

  2. 02
    Trace

    Follow code, state, permissions, persistence, and tests.

  3. 03
    Reproduce

    Separate the PR regression from an existing defect.

  4. 04
    Probe

    Test the smallest safe fix in isolation.

Across the stack

Different tools. The same care.

Linux infrastructure, speech tools, ML pipelines, native apps, and Godot games. In each, I care about the whole path from a working experiment to something another person can run and continue.

Lab bench

Curiosity down to the circuit.

On the lab bench: digital logic, a working 4-bit CPU with a custom instruction set in Turing Complete, and IC10 automation. Independent experiments in how machines work.

04 / Contact

Have something worth building?

Get in touch about systems, security, applied AI, or a prototype that needs a working path forward.

quick.nav

Where do you want to go?

Tip: press Ctrl + K from anywhere.