AI Engineer
Combining advanced AI engineering with fullstack development and work psychology to build systems that don't just work — they learn and support human growth.
In the Rhythm of Code
0
—
—
0
—
∞
Brewed in CachyOS, Wrocław, and a dozen cafés across Lower Silesia.
Psychology ×
engineering,
rendered in code.
I am not just a typical developer. My background in psychology allows me to view code through a human lens. I understand not just HOW to build something, but fundamentally WHY and FOR WHOM we are building it.
Code is a behavioural artefact — it shapes attention, time and trust.
AI Engineering
Multi-agent systems (Hermes-style orchestration, LangChain, autogen), OpenAI API integration (GPT-4, embeddings, function calling), RAG pipelines with pgvector, prompt engineering. Production deployments for clients since 2023.
Fullstack Development
Next.js 16, React 19, TypeScript 5.7, Tailwind CSS v4. Backend: Node.js, PHP, WordPress REST API, PostgreSQL, Supabase, Redis. Headless CMS, e-commerce, SaaS — from idea to production.
Work Psychology
Self-Determination Theory (Ryan & Deci), Herzberg two-factor model, habit loops, organizational design. I design systems for intrinsic motivation, not dark patterns.
UI/UX & Automation
Design systems, Framer Motion, accessibility (WCAG 2.2 AA), CI/CD, GitHub Actions, Vercel. Workflow automation: Make, n8n, custom scripts connecting AI with existing infrastructure.
My Process
I don't write code in a vacuum. Every line stems from a deep understanding of the problem and the human experiencing it.
Psychology & Analysis
I start with 'Why?'. I analyze user motivations (SDT), habit loops, and business goals. I define the problem before solving it.
Psychology & Analysis
I start with 'Why?'. I analyze user motivations (SDT), habit loops, and business goals. I define the problem before solving it.
Strategy & Architecture
I choose technology for the problem, not vice versa. I plan data structures, information flows, and security. I create 'Blueprints'.
Strategy & Architecture
I choose technology for the problem, not vice versa. I plan data structures, information flows, and security. I create 'Blueprints'.
Clean Code & Execution
I write code that is readable by humans and efficient for machines. I prioritize Type Safety, automated tests, and accessibility (WCAG).
Clean Code & Execution
I write code that is readable by humans and efficient for machines. I prioritize Type Safety, automated tests, and accessibility (WCAG).
Verification & Evolution
Deployment is just the beginning. I analyze metrics, gather feedback, and iterate. The product lives and evolves with its users.
Verification & Evolution
Deployment is just the beginning. I analyze metrics, gather feedback, and iterate. The product lives and evolves with its users.
Selected Projects
Technology makes sense only when it solves real problems. Here are a few examples of how I translate code into value.
Wrocław Lead Gen Platform
AI-powered B2B lead acquisition platform: Apify scraping for businesses without websites, AI landing page generation (Claude/Cursor/v0), 990 PLN/site sales.
- Apify
- Claude API
- React + Vite
- AI Lead Gen
Built where
agents meet
production.
Real AI implementations in client and personal projects — from multi-agent systems to RAG pipelines.
multiAgent
Orchestration of AI agents (Claude/Hermes-style) with role-based tool calling, function routing, and result-driven control flow. Architectures where every agent has a narrow specialization and a clear contract.
agent-router → researcher → critic → writerllmIntegration
Integrations with OpenAI API (GPT-4, embeddings, function calling) and Anthropic. Streaming, retries, rate limiting, structured outputs. Production systems, not demos.
POST /v1/chat/completions → parse → tool dispatchpromptEngineering
System prompts written like code: versioned, tested, evaluated. Function calling schemas, JSON-mode, prompt caching, A/B evals. Zero 'magic prompts' — everything in the repo.
prompts/v2/gpt-4o-motivation-analyzer.tsragPipelines
Retrieval-Augmented Generation: chunking, embeddings (text-embedding-3-small), vector store (pgvector / Pinecone), re-ranking, context window optimization. End-to-end with own vector database.
query → embed → retrieve k=20 → rerank top 5 → LLMLet's talk about
Your Project
Looking for a fullstack developer who understands user psychology? Or maybe you want to consult on an employee well-being app idea? Get in touch.
Location
Poland / Remote