Hi — I'm Laura, a software engineer building AI-native applications and platforms.
I specialize in designing and implementing production-grade AI systems — from agent orchestration and retrieval pipelines to full-stack interfaces and deployment workflows.
- 🧠 AI systems engineering — LLM orchestration, RAG pipelines, decision-support platforms
- 🏗 Full-stack architecture — TypeScript, Next.js, Node.js, monorepos, testing & CI/CD
- ☁️ Production mindset — reliability, observability, evaluation, and scalable system design
- 🎓 Bachelor's Degree in Business Informatics
- 💼 8+ years building complex applications in collaborative environments
- 🌍 Based in Germany — open to international opportunities
Currently focused on building real-world AI applications that operate reliably outside the lab.
- Agent orchestration & tool-calling workflows
- Retrieval-Augmented Generation (RAG)
- Structured evaluation & testing of LLM systems
- Prompt engineering & guardrails
- Decision-support system design
- AI system observability & quality monitoring
- Human-in-the-loop interfaces
|
Mosel Flood Risk Intelligence Platform
A production-style AI decision-support platform combining live sensor ingestion, rule-based risk scoring, retrieval-augmented evidence, and LLM-generated explanations — delivered through a real-time Next.js dashboard. 
|
|
Financial Document Intelligence Agent
A RAG-powered Q&A system for financial documents — upload a 10-K or earnings report, extract key metrics as structured JSON (revenue, EPS, net income, YoY deltas), then ask natural language questions and get grounded answers with inline citations. 
|
Flood Intelligence — architecture highlights
- Real-time environmental data ingestion
- Rule-based risk modeling
- Retrieval-Augmented Generation for explainability
- LLM-driven decision interpretation
- End-to-end full-stack system (API → orchestration → UI)
- Monorepo architecture with automated CI and quality gates
Stack: TypeScript · Next.js · Node.js · pnpm monorepo · Claude AI (Anthropic) · Vitest · SonarCloud
Financial Document Intelligence — architecture highlights
- Section-aware PDF chunking (SEC 10-K Item detection)
- ChromaDB vector store with document-scoped similarity search
- Structured metric extraction with JSON schema prompting (revenue, EPS, gross margin, guidance)
- RAG pipeline with grounded answers and inline citations
- FastAPI backend + Next.js App Router frontend with server-side proxy
- Multi-stage Docker builds with ONNX embedding model baked in
Stack: Python · FastAPI · Next.js · ChromaDB · fastembed ONNX · Claude AI (Anthropic) · pytest · SonarCloud
- AI orchestration platforms for real-world decision support
- Evaluation frameworks for reliable LLM workflows
- End-to-end AI product architectures
AI platform architecture · distributed systems · developer tooling · system reliability · human-AI interaction
AI platform engineering · agent systems · production LLM infrastructure · intelligent decision-support software
I enjoy exploring new technologies, building real systems, and continuously improving how intelligent software operates in production environments.