Full-Stack Developer | React · TypeScript · FastAPI · AI Tools
I build safety-first AI applications — from typed APIs and automated tests to responsive web, mobile and desktop interfaces.
My work centers on full-stack delivery, Python/FastAPI development, AI tooling and automation, and safety-first product engineering.
Available now for remote B2B, freelance and contract work.
Location: Poland
Email: mateusz.ozimkiewicz9@gmail.com
MellyTrade is a safety-first, read-only AI trading terminal and paper-risk workspace. It combines a Python and FastAPI backend with a React, TypeScript and Vite frontend, typed Pydantic schemas, a mobile/PWA route, and a Tauri desktop thin shell.
The repository provides verifiable engineering evidence through pytest safety invariants, OpenAPI forbidden-path tests, Playwright end-to-end coverage, Docker, GitHub Actions, and public demo infrastructure on Render and Vercel. The interface focuses on market context, broker status, portfolio and risk views, paper planning, and audit-friendly safety signals.
The public demo is deliberately constrained: it is read-only, dry-run, and paper-only where applicable. Live orders are blocked, human review is required, maximum risk is capped at 1% per trade, and no real-money execution is available. The project demonstrates safety-oriented product and engineering decisions; it does not claim profitability, verified returns, or commercial trading performance.
MellyCore AIOS is a static-first AI command center and Living Context Graph foundation for coordinating agents, context, specifications, and safety gates. It explores how documentation, structured evidence, and graph relationships can make complex AI-assisted work easier to inspect and hand off.
Verified repository features include the Living Context Graph, Knowledge Graph Console, multi-agent context handoff, docs-first product architecture, safety-aware design rules, a static responsive showcase, and visual QA evidence. The current public scope has no backend or provider integration, no database, no live ingestion, and no secrets. It is an architectural and interface foundation rather than a claim of a live AI runtime, real-time ingestion system, or autonomous production agent platform.
- Backend: Python · FastAPI · Pydantic · pytest
- Frontend: React · TypeScript · Vite · PWA
- Desktop: Tauri
- Quality and delivery: Playwright · Docker · GitHub Actions · Git
- AI workflow: LLM tools · agent workflows · context engineering · structured documentation
AI tools support research, iteration, code review, testing, and documentation. I use them to explore options and shorten feedback loops, while keeping the work reviewable through source control, tests, explicit constraints, and written evidence. Architecture, safety constraints, validation, and final engineering decisions remain human-owned.
- Building clear full-stack portfolio case studies.
- Improving MellyTrade's public engineering evidence.
- Developing MellyCore's Living Context Graph.
- Open to Full-Stack, Python and AI tooling opportunities.
-
Location: Poland
-
Work mode: Remote
-
Availability: Immediate
-
Engagement: B2B, freelance or contract
-
Portfolio: mateusz-ozimkiewicz-portfolio.vercel.app