17.06.2026 aktualisiert
100 % verfügbar
Senior AI/ML Engineer | Computer Vision | LLM/RAG | Medical AI | MLOps & AI Governance
Heidelberg, Deutschland
Weltweit
Ph.D. in Computer Science (magna cum laude), University of TübingenÜber mich
Senior AI/ML Engineer and Applied Scientist with 10+ years of experience in Computer Vision, LLM/RAG systems, Medical AI and MLOps. I support companies from AI concept and prototyping to model development, evaluation, reproducible pipelines and AI governance/EU AI Act readiness.
Skills
RAGLLMKünstliche IntelligenzComputer VisionMedizinische BildgebungMachine LearningAzure Machine LearningDeep LearningMachine Learning Operations
AI/ML & Computer Vision
- I design, train, evaluate and document ML/deep learning models for real-world image, video and multimodal data.
- Experience includes object detection, classification, segmentation, video understanding, human pose estimation, pedestrian detection, medical image analysis and computational pathology.
- Keywords: PyTorch, Computer Vision, Deep Learning, Vision Transformers, Self-Supervised Learning, Vision-Language Models, Object Detection, Segmentation, Video Analysis.
LLM, NLP & Agentic Systems
- I build LLM applications, RAG pipelines and AI-assisted knowledge systems for domain-specific use cases.
- Work includes document processing, semantic search, prompt engineering, information extraction, evaluation workflows and multi-agent systems.
- Keywords: LLMs, RAG, LangChain, LangGraph, LlamaIndex, OpenAI API, Anthropic API, NLP, Semantic Search, Multi-Agent Systems.
MLOps, Cloud & Reproducibility
- I build reproducible, maintainable and deployment-oriented AI workflows with clear tracking, documentation and handover.
- I help teams structure ML projects, containerize workflows, improve evaluation and prepare models for reliable use.
- Keywords: Docker, MLflow, Weights & Biases, AWS, Azure ML, CI/CD, Experiment Tracking, Containerized Workflows, Reproducible Pipelines.
Programming & Data Engineering
- I develop clean, modular Python code for AI/ML systems, data pipelines and research-to-application workflows.
- I work with large-scale scientific, medical and multimodal datasets, including image, video, text and structured data.
- Keywords: Python, OOP, Clean Code, C/C++, CUDA, Bash, Git, Testing, REST APIs, HDF5, PostgreSQL, MongoDB.
AI Governance & Compliance
- As a TÜV-certified AI Officer, I support organizations in building, documenting and governing AI systems responsibly.
- I help with risk classification, documentation, validation concepts, audit trails and governance for regulated or high-risk AI use cases.
- Keywords: AI Governance, KI-Beauftragter, EU AI Act, ISO/IEC 42001, GDPR, Risk Classification, AI Documentation, Validation, Audit Trails.
Medical AI & Scientific AI
- Specialized experience in medical AI, computational pathology, spatial transcriptomics and clinical decision-support prototypes.
- Work includes cell segmentation, cell-type annotation, gene-expression prediction, surgical video understanding and clinical AI projects.
- Keywords: Medical AI, Computational Pathology, Spatial Transcriptomics, Cell Segmentation, Surgical Video Understanding, Clinical Decision Support.
Foundation Models
- I develop and evaluate advanced deep learning approaches for visual, multimodal and domain-specific AI tasks.
- Keywords: Foundation Models, Vision Transformers, Self-Supervised Learning, Vision-Language Models, Representation Learning, Multimodal AI.
Erfahrung
Berufserfahrungmehr als 10 Jahre
Freelancer seit2026
Sprachen
DeutschgutEnglischverhandlungssicherTürkischMuttersprache
Projekthistorie
Developed reproducible Python pipelines for high-throughput Computational Pathology and spatial transcriptomics (Visium HD), including cell segmentation, cell-type annotation, and gene-expression prediction in containerized workflows. Built a RAG/LLM pipeline to standardize approximately 16,000 free-text surgical descriptions to SNOMED CT. Developed and evaluated foundation models for surgical video understanding using Vision Transformers, self-supervised learning, and Vision-Language approaches.
Led an AI project on facial phenotyping for rare genetic disorders, improving clinical experts diagnostic accuracy by 20%, built a clinical decision-support prototype, and coordinated validation with the National Institutes of Health (NIH, USA). Conducted additional research on 3D face shape modeling for emotion and action-unit recognition.
Developed AI models for automated student engagement estimation using multimodal real-world classroom data. Designed and built a multi-camera IP network for real-time audiovisual data acquisition, led data collection at partner schools, drafted the ethics application, and implemented data-protection requirements.