Skip to content
View justi's full-sized avatar
74 followers · 116 following

Achievements

Achievement: Pair Extraordinairex4Achievement: QuickdrawAchievement: StarstruckAchievement: YOLOAchievement: Pull Sharkx3Achievement: Arctic Code Vault Contributor

Achievements

Achievement: Pair Extraordinairex4Achievement: QuickdrawAchievement: StarstruckAchievement: YOLOAchievement: Pull Sharkx3Achievement: Arctic Code Vault Contributor

Organizations

@spree@Okazjum@GetBadges@EnviroMonitor

Block or report justi

Block user

Prevent this user from interacting with your repositories and sending you notifications. Learn more about blocking users.

You must be logged in to block users.

Maximum 250 characters. Please don’t include any personal information such as legal names or email addresses. Markdown is supported. This note will only be visible to you.
Report abuse

Contact GitHub support about this user’s behavior. Learn more about reporting abuse.

Report abuse
justi/README.md

Hi, I'm Justyna 👋

📍 Poznań, Poland | 💎 Ruby on Rails Engineer | 🚀 Building production apps

RubyRailsJavaScriptPythonPostgreSQLpgvectorHotwireTailwindCodexClaudeOllamaOpenCodemacOSWeb

Full-stack Rails developer shipping production apps - from health tools and EdTech to SaaS platforms and marketing automation. AI-assisted, human-driven.

Production Projects

  • 🛒 Snipe Sale 🇬🇧🇵🇱🇫🇷🇩🇪🇪🇸🇮🇹 - Track promotions from niche shops, not products - 6 country markets
  • 🥑 KetoLabs.pl 🇵🇱 - Keto/LCHF/LCHP education hub - built it because I follow LCHF myself
  • ✍️ OutlineShare 🇬🇧 - LinkedIn content workflow: write, schedule, publish, grow
  • 📝 CramZap 🌍 / Odpytywarka.pl 🇵🇱 - Upload a PDF, get an interactive quiz (OpenAI-powered) - 11 language versions
  • 🔍 HybridRanking 🇬🇧 - SEO auditing tool with real-time analysis
  • 📚 Conftrace 🇬🇧 - Research papers from 36 top CS/AI conferences: search, authors, topics, trends - built on my researchpooler fork (of karpathy's)
  • 🎭 PersonaTool 🇬🇧 - Product-market fit testing with LLM-generated synthetic personas
  • 📖 Anglofaza 🇵🇱 - Learn English with fresh BBC articles (OpenAI-powered) - my first AI-driven project (2023)

Open Source

  • 📜 ruby_llm-contract - Validate and retry LLM outputs for ruby_llm: describe the expected JSON, fall back to a stronger model, gate CI on regressions (demo)
  • 🖥️ Aya - Desktop workspace for long-lived coding-agent terminals (Claude Code, Codex, Aider, ...) - co-author, with @khasinski
  • 🧭 ollama-task-router - Task-based prompt router for Ollama: a small model classifies each prompt and routes it to the best local specialist
  • ⚡ ollama-bench - Reproducible, zero-dependency benchmark of local LLMs: speed (tok/s), energy, reasoning and code quality
  • 🔁 ollama-agentic-loop - Benchmark of local agentic coding: test-feedback loop over Ollama + solve@K scorer
  • 🔒 claude-code-project-boundary - Claude Code plugin: scope-aware protection for destructive ops within project boundaries
  • 💪 stimulus-password-strength - Importmap-friendly Rails 8 password strength field with Stimulus & zxcvbn
  • 💰 PriceScanner - Ruby gem for multi-currency price extraction from text
  • 🧹 no_comments - Ruby gem that strips unnecessary comments from .rb files, leaving code clean for deployment
  • 📷 xiaomi-cw300-unifi - Bridge a cloud-locked Xiaomi CW300 camera into UniFi Protect via go2rtc + a minimal ONVIF adapter
  • 🔗 Contributor to langchainrb - merged fix for Chroma vector search queries (#890, released in 0.19.3)
  • 🔭 Contributor to MulensModel - gravitational microlensing modeling in Python: 13 merged PRs (CI setup, bug fixes, test coverage)
  • 📊 Contributor to ccusage - co-authored fix: honor --since/--until in the OpenCode SQLite loader (#1492)

What I'm About

  • Rails 8 all the way - Hotwire, Turbo, Stimulus, no heavy JS frameworks needed
  • AI-augmented development - Claude Code (agents, skills, commands, hooks) and OpenAI to ship faster
  • Local LLMs too - benchmarking and routing local models on Ollama, measured, not guessed
  • Ship it - from idea to production in days, not months
  • Solo founder mode - designing, building, deploying, and marketing entire products

Connect

TwitterLinkedInWebsiteGitHub

I architect, direct UX, and orchestrate AI agents - they code, I steer. Powered by Polish coffee ☕

Pinned Loading

  1. claude-code-project-boundaryclaude-code-project-boundaryPublic

    Claude Code plugin: scope-aware protection — allows destructive ops (rm, mv, chmod) within project, blocks them outside. Built for dangerouslySkipPermissions mode.

    Shell 86 3

  2. ruby_llm-contractruby_llm-contractPublic

    Validate and retry LLM outputs for ruby_llm. Describe the JSON response you expect, fall back to a stronger model when the cheaper one fails the rules, and gate CI on regressions — all as one contr…

    Ruby 35

  3. khasinski/ayakhasinski/ayaPublic

    Tabbed terminal manager for coding agents (Claude Code, Codex, shell). Real PTYs, project-grouped tabs, search-everything, theme import.

    TypeScript 18 3

  4. ruby_llm-contract_demoruby_llm-contract_demoPublic

    Companion demo for ruby_llm-contract showing the LLM-as-judge pattern: build → calibrate → gate. Polish e-commerce chatbot scenario, live OpenAI calls (gpt-4.1-mini).

    Ruby 3

  5. ollama-task-routerollama-task-routerPublic

    Task-based prompt router for Ollama: a small model classifies each prompt and routes it to the best of three local specialists (coder / reasoner / quick).

    Python 2

  6. ollama-benchollama-benchPublic

    Reproducible, zero-dependency benchmark of local LLMs on Ollama: speed (tok/s), energy, reasoning and code quality, driven by a single-source-of-truth config pipeline.

    Python 2