Full Stack Software Engineer
Based in Brandenburg, Germany
Get in Contact
Strong infrastructure starts small.
One commit at a time.
Until it reaches millions.
About / Person
Full Stack Software Engineer / Brandenburg, Germany
I build software that stays reliable when usage, complexity, and expectations grow. I care about clean architecture, clear communication, and shipping with intention.
Outside of coding, I recharge through movement, gaming, and especially (psycho) thriller movies.
Motivation
My motivation comes from seeing people use the product I have designed and built. I enjoy taking messy problems, reducing them to understandable parts, and delivering solutions that actually solve the problem. Coding without impact is like playing a game without a goal: it can be fun, but it doesn't feel meaningful.
Soft Skills
How I Work
I work best in transparent loops: understand goals, align early, ship small increments, and iterate based on signal. I value pragmatic decisions that preserve long-term maintainability.
Personality
My Working Tree
Education, enterprise delivery and founding a company overlap. Scroll through the branches to see what each one involves.
Active branch
01/03Formal grounding, carried on in parallel with everything else.
2019 — Present
Track log
4 nodes
2019 - 2022
Built the first foundation in structured learning and problem solving.
2022 - 2025
Combined academic depth with practical execution in parallel environments @THB with ZF.
2025
Research on context engineering and designing context-aware AI Systems.
2025 - Present
Pursuing my master's degree in parallel with startup and enterprise responsibilities @THB.
Tech Stack
A quick view of the technologies I use most often to build reliable products, from interface architecture to backend delivery and operations.
Languages & Frameworks
04My default language for production work. It keeps contracts explicit, refactors safe, and large codebases maintainable.
Used inHumaifyLibrary OccupancyChatbot CharlotteJson-Formatter
For full-stack product delivery, I use Next.js to unify routing, server rendering, and performance-sensitive page architecture.
The component model I think in. I lean on it for predictable state boundaries and interfaces that stay debuggable as they grow.
Used inLibrary OccupancyInvoice App
I rely on Node for API platforms and backend services that need a clean dev loop, solid observability, and stable runtime behavior.
Used inInvoice App
Interface
02For UI execution speed, Tailwind lets me craft precise systems quickly while preserving strong visual structure and consistency.
Used inInvoice App
Where a tool has to run offline and own its own window, Electron lets me ship the same interface as a desktop application.
Used inInvoice App
Data & AI
03I design schemas and queries for reliability first, with indexing and migration discipline that scales as product complexity grows.
Used inHumaifyLibrary Occupancy
I use vector search where semantic retrieval matters, with configurations tuned for practical latency and predictable recall.
Used inHumaify
For retrieval-augmented systems, it gives me the indexing and query layer that turns scattered internal documents into answerable context.
Used inChatbot Charlotte
Infrastructure
03I use containerized workflows to keep environments reproducible and deployment pipelines consistent from development to production.
Used inJson-Formatter
Where a service should be deployed without standing up infrastructure first, Railway gets it running with sane defaults and clear logs.
Used inHumaify
For private access between machines and services, it removes the usual VPN overhead while keeping the network boundary explicit.
Used inLibrary Occupancy
Selected Projects
Practical products across edge sensing, analytics, AI workflows, data quality, and developer experience.
AI voice Agents that make an Impact.
Agentic Platform
01Humaify bridges AI with real-world applications through a core competency in voice agent platforms. As lead developer, I architected and built a scalable platform featuring a Next.js frontend, Node.js backend, and Qdrant vector database for context management. The Monorepo is engineered to support rapid development and deployment of custom voice agents across diverse use cases.
Conversational Context Aware AI
A context-aware assistant for internal workflows.
A context-aware support assistant that addresses recurring questions, surfaces process knowledge, and escalates complex issues with full conversation history. Built on LLama Index to index and query internal documentation, delivering accurate contextual responses to employee requests.
Developer Tooling
Fast formatting and validation for JSON payloads.
A lightweight JSON formatter with syntax validation and readability presets for streamlined API debugging and payload inspection. Available as a containerized Docker image.
IoT & Edge AI
02A privacy-preserving occupancy system for the university library at TH Brandenburg, built with a six-person team where I owned the technical concept and architecture. Bluetooth paxcounting and cameras were both rejected — one too imprecise, the other a data-protection problem — for a ceiling-mounted Hesai JT16 Mini LiDAR on an Nvidia Jetson Orin Nano. A Rust service clusters the raw point cloud and counts entries and exits through configurable zone polygons, so people register as shapes and are never identified. Counts sync over Tailscale to a Next.js dashboard on PostgreSQL, with a monthly-retrained LSTM driving hourly forecasts. Counting landed near 3.6% average deviation; the forecast is still data-limited.
Business Application
An invoice management system for small businesses.
A comprehensive invoice management system built with React and Node.js, enabling users to create, edit, track invoices, and generate reports. Packaged as an Electron desktop application for offline-first operation without external infrastructure. Built to help my father efficiently manage his small business invoicing while providing a free solution for other entrepreneurs. Releases for both Windows and Mac are available on the GitHub repository
Duplicate Finder
Detecting duplicates
A duplicate file detector using hash-based comparison across directories. Generates content-based hashes and stores them in a HashMap for O(1) lookup, enabling efficient duplicate identification.
Data Analytics
03Developed an ML model trained on customer behavior, apartment, and environmental data to predict rental duration and forecast apartment vacancy timing.
