Profile  ·  Current systems  ·  Agentic development  ·  Stack  ·  Public work

I am a dual Business Informatics student and software developer working where product thinking, architecture, implementation, and operations meet. I started programming at 12 and have since worked across scripting, full-stack products, automation, native clients, developer tooling, and game systems.

I prefer building complete, understandable systems rather than isolated features. A change should remain traceable from the user need through architecture, code, tests, documentation, deployment, and operation. The same principle shapes my game work: narrative and visual ideas must survive contact with gameplay, pacing, scene logic, animation, and engine constraints.

What I focus on

• Production web platforms with frontend, backend, data, authentication, administration, and operations designed together.
• Agentic engineering workflows with specialist roles, delegated sub-agents, structured handoffs, scheduled automation, and independent QA.
• Unity game architecture, narrative implementation, visual-development pipelines, reusable packages, and supporting web services.
• Delivery systems that are observable, reversible, documented, and proportional to the actual project risk.

Most of my active engineering work lives in private repositories. The public repositories below are a selected cross-section rather than the complete footprint.

A production-oriented vehicle sourcing and decision platform for dealer workflows. The system spans listing acquisition, normalization, structured analysis, recommendations, multi-tenant accounts, administration, native packaging, deployment, observability, and role-specific documentation.

Angular · TypeScript · Spring Boot · Java · PostgreSQL · Tauri · Rust · Docker · nginx

An original, story-driven 2D game universe where charm meets dread. My work across the project connects narrative design, playable scene implementation, Unity systems, visual development, shared tooling, web infrastructure, and production coordination.

Unity 6 · C# · URP 2D · Java · Angular · Narrative systems · Art pipelines

_{Original SolutionWCMD key art from the project visual-development library.}

GitHub organization  ·  Project website  ·  YouTube

I use AI-assisted development as an engineering operating model, not as a single chat or autocomplete layer. Specialized agents receive explicit scope, repository context, ownership, constraints, and acceptance criteria. They can delegate bounded work to sub-agents, exchange structured handoffs, run on schedules, and move changes through review, CI, documentation, and release checks.

Human direction remains responsible for architecture, priorities, trade-offs, risk acceptance, and production decisions.

The practical techniques behind the model include:

• Agent-to-agent delegation: an orchestrator or specialist creates narrowly scoped sub-agents when parallel work or dedicated expertise improves the result.
• Structured communication: agents exchange repository artifacts, issues, pull requests, review notes, decision records, and explicit handoff summaries instead of relying on implicit session memory.
• Scheduled workflows: recurring development, repository hygiene, documentation, QA, and CI routines run automatically at defined times or events.
• Independent verification: feature creation and quality approval remain separate responsibilities, with deterministic checks and human-controlled release gates.

Applications and data

Interactive and native systems

Delivery and operations

Engineering practices

System architecture · Agent orchestration · Automated QA · Release engineering · Documentation as code · Observability · Narrative implementation

1. Architecture should reduce future decisions, not create ceremony.
2. Automation should remove repeated effort while keeping responsibility visible.
3. QA, documentation, and operations belong inside development, not after it.
4. Release strictness should match project risk, team size, budget, and recovery options.
5. A narrative script is only successful when it also works as playable content.