Backend / Fullstack Developer · AI Character Systems
I build systems where code, behavior logic, memory, perception and long-term interaction meet.
🟢 Open to junior Backend / AI-integration roles · remote
- 💻 Backend, fullstack, AI systems and automation
- 🛠️ I like projects that combine code, behavior logic and system architecture
- 🤖 Focused on local AI characters, LLM pipelines and character-based runtimes
- 📚 Improving OOP, testing, clean backend architecture and project structure
- ⚙️ I use AI tools as accelerators, but I shape and review the architecture myself
M.A.R.I.A. is my long-term author-driven local AI character ecosystem — not a chatbot or a companion SaaS, but a believable single-user character with memory, subjective perception, personality and long-term continuity, running locally through Ollama.
Under the hood: a self-written ML perception brain, a cognitive memory / dossier layer, a dedicated local world-runtime and a permission-gated controlled-agency layer.
👉 Explore the ecosystem through the hub — M.A.R.I.A.
The hub is the public entry point — philosophy, ecosystem map and direction. A sanitized runtime
snapshot lives in M.A.R.I.A.-Core-Sanitized-Snapshot; active development continues privately.
Role-based access (admin / teacher / student) with modules, lessons, tests and grades. Split into two repositories:
- online_platform_api · Backend ·
PythonFastAPI + async SQLAlchemy on PostgreSQL · JWT auth with login throttling · dependency-based role guards · management of users, groups (Excel import), modules, subjects, lessons, tests and grades. - online_platform · Frontend ·
JavaScriptReact + Vite · authentication & protected routes · student personal account · module/lesson pages · tests & grades · teacher and admin panels.
🧪 Sandbox_with_LLM — LLM experiments
A branch-based sandbox for small LLM-powered systems — each branch is an independent prototype with its own goal, README and tests.
📨 Telegram_Bot — NSTU Helper · Python
Telegram bot for NSTU freshmen (built on aiogram 3): RU/EN navigation, campus events, popular places and quick surveys.
- Growing M.A.R.I.A. as a single-user local AI character runtime — perception, memory, behavior
- Cleaner Core / external-clients architecture and API contracts
- Stronger backend design, OOP, tests and project structure
- Local LLM pipelines, memory systems and character-based agents
- GitHub: @HontoUKI
This profile is part of my path as a developer, system thinker and creator of M.A.R.I.A.