23.10.2026 aktualisiert
Premiumkunde
100 % verfügbarAI Engineer & Backend Developer | LLM & RAG Systems
Berlin - Schöneberg, Deutschland
Deutschland
M. Sc. Cognitive SystemsÜber mich
Ich bin AI Engineer und Backend Developer mit fundierter Erfahrung in der Entwicklung von LLM-Systemen, Automatisierungs-Pipelines und skalierbaren Backend-Architekturen. Mein Schwerpunkt liegt auf Generativer KI, RAG, human-in-the-loop Systemen und moderner Python-Entwicklung.
Skills
AI AgentsAzure KubernetesAzure Machine LearningLarge Language ModelsMulti-Agent SystemsGenerative AIBackendBackend-programmierung
As an experienced AI Engineer, I combine deep AI research with robust software engineering. I don't just build mere demo projects, but scalable, production-ready backend architectures and autonomous AI systems that automate real business processes and create measurable added value.
Why work with me?
End-to-End Expertise: From concept and backend implementation (Python, FastAPI) to LLM fine-tuning, deployment, and observability (vLLM, Langfuse, OpenTelemetry), I deliver clean, test-driven software.
Focus on Agentic Workflows & Automation: I help companies automate complex manual processes (such as data migrations or claims management) through smart multi-agent systems and LLM integrations.
Scientifically Grounded, Pragmatically Implemented: Thanks to my background in cognitive systems, I understand not only the APIs, but also the underlying mechanics (RAG, Fine-Tuning, Knowledge Graphs).
Technical Expertise
AI & LLM Engineering:
- Agentic Systems & Frameworks: LangGraph, Multi-Agent Architectures
- LLM Optimization & Deployment: vLLM, llama.cpp, Unsloth (LoRA Fine-Tuning), Prompt Experimentation & Versioning (Langfuse)
- Core & ML: PyTorch, Hugging Face, scikit-learn, Pandas
Backend & Databases:
- Python, FastAPI
- PostgreSQL, SQLAlchemy, Alembic, Qdrant & Pgvector, Neo4j, OpenMetadata
Cloud & DevOps:
- Azure (Functions, Logic Apps, ML Pipelines)
- CI/CD, Kubernetes, OpenTelemetry, Arize Phoenix, Sentry
Software Practices:
- Test-Driven Development (TDD), Agile Mindset, Modern AI-Assisted Workflows across the entire SDLC
Industries & Use Cases
- Regulated Industries & Insurance: Automated, traceable data migration systems with LLM agents, confidence scoring, and human-in-the-loop validation.
- Customer Operations & Claims Management: AI-driven systems for fully automated processing of customer inquiries (from email intake to autonomous response generation).
- Medical Technology & IoT: Development of voice- and AI-supported interfaces (e.g., smartwatch chatbots for medical questionnaires via Rasa, LangGraph, Pipecat).
- Data Engineering & Knowledge Management: Building knowledge graphs and semantic search systems from unstructured business documents.
Looking for a strong partner to implement your AI or backend projects? Let’s talk—I look forward to your message!
Erfahrung
Berufserfahrung5 - 10 Jahre
Freelancer seit2026
Sprachen
DeutschMutterspracheEnglischverhandlungssicher
Projekthistorie
AI Engineer in a data and AI consultancy delivering projects for clients in regulated industries. Building an AI-supported data migration system for an insurance client using LLM agents that propose mappings and SQL transformations from a legacy database to a new target schema. Leveraging schema metadata, data profiles, and a knowledge graph built from business documentation (Neo4j, OpenMetadata, Harness Engineering). Proposed mappings are validated, scored for confidence, and traceable, with domain expert review and feedback loops. Observability implemented with OpenTelemetry and Arize Phoenix.
Developed an AI-driven Customer Claims Management system using Python and Azure Functions, automating the full lifecycle from mail extraction to autonomous response generation. Designed a continuous learning pipeline using Unsloth for LoRA adapter finetuning and vLLM for high-throughput deployment, managed via a human-in-the-loop system with Langfuse for human annotation, labeling, and prompt versioning. Created a smart watch chatbot for medical questionnaires with Python, Rasa open source and LLM support, with migration to LangGraph and Pipecat. Collaborated closely with backend, frontend, and DevOps engineers.
Developed a vision transformer to autonomously identify changing point sources in X-Ray telescope image time series (Master's thesis). Built a binary image classifier to differentiate extended and point-like X-Ray sources. Managed databases and websites using Python, Flask, PostgreSQL, and Docker. Created an advanced repository for cross-matching astronomical surveys, generating light curves, visualizing results in PDF formats, and integrating automated web scraping with Selenium.