To support the digital transformation of legal domains, C.H. Beck initiated the development of scalable machine learning solutions, particularly in the area of Natural Language Processing (NLP), with the goal of automating repetitive manual tasks and enabling new digital products. I played a leading role in the technical implementation and independently handled the following tasks:

• Analysis and prioritization of business requirements in close collaboration with product management
• Development of production-ready NLP applications based on MLOps principles, focusing on Named Entity Recognition (NER)
• Design and implementation of a scalable, multi-stage infrastructure for data preparation, model development, and deployment in Azure using Azure Machine Learning
• Planning and execution of Proof of Concepts (PoCs) with Generative AI and LLMs (LangChain, Azure OpenAI, pydantic-ai) to assess innovative application areas
• Creation of project plans with clear milestones, resource planning, and realistic effort estimations
• Maintenance and further development of existing C#/.NET applications, including feature implementation, code reviews, test automation, quality assurance, as well as support for deployment and delivery

Results & Impact:

• Successful operationalization of an NER use case for legal document analysis
• Increased flexibility compared to the previous regex-based solution
• Reduction of manual workload in data processing
• Establishment of a modular ML infrastructure as a technological foundation for future LLM-based applications

Local clubs, initiatives, and social organizations are often difficult to find because information is scattered across websites, PDFs, or social media. A central platform that makes social places, activities, and engagement opportunities visible and connected did not previously exist. The goal is to develop a digital platform that automatically collects, structures, and makes social places, organizations, and events easily accessible to citizens. The focus is on a modern, cloud-based architecture, high automation, and the use of agentic AI to intelligently process and curate content.

• Architecture & Infrastructure: Built a scalable cloud environment on Azure using Docker, Terraform, Azure App Service, Azure Container Instances, and KeyVault.
• Backend Development: Developed a modular FastAPI application with PostgreSQL, SQLAlchemy, and Alembic; designed ETL pipelines for automated data integration.
• Automation & CI/CD: Implemented a GitLab CI pipeline with automated testing, deployment, and semantic release processes; ensured code quality with Ruff, Mypy, Pre-commit, and Pytest-Cov.
• Agentic AI & Data Processing: Leveraged GPT and LLaMA models via Azure AI Foundry for automated text processing, classification, and generation of structured descriptions; developed autonomous agents to analyze data sources, summarize content, and provide contextual insights.
• Web Crawling & Data Integration: Built a web crawler using BeautifulSoup, Pandas, and Hydra-Core to extract, clean, and standardize data on clubs and events.
• Frontend & User Experience: Developed a modern React interface using Chakra UI, Leaflet for interactive maps, and Vite for high-performance rendering; implemented multilingual support (i18n).
• Governance & Product Development: Defined requirements, roadmap, and brand identity; responsible for architecture, development, and product strategy.
• Monitoring & Operations: Implemented monitoring and logging components to track performance, availability, and data quality.
• Utilized GitHub Copilot to increase development efficiency and integrate AI-powered tools into existing workflows.

Result

It is under active development. Even at this stage, the platform makes social places and opportunities visible, strengthens local communities, and provides the technical foundation for future expansion to additional cities and regions.

In my role as Senior Machine Learning Engineer at adesso SE, I supported clients across various industries in developing and operationalizing data-driven solutions. My focus was on building scalable AI infrastructures in the cloud, developing production-grade ML applications, and establishing modern MLOps practices. I held roles such as Lead ML Developer, AI Architect, Proxy Product Owner, and Deputy Project Lead.

Project:

Scalable ML Infrastructures for Cloud Platforms (Azure, AWS, MLOps)

Companies in healthcare, aviation, energy, and manufacturing aimed to automate and optimize processes using data-driven approaches but often lacked suitable ML infrastructures or development standards.

• Built cloud-based ML platforms in Azure and AWS, including CI/CD, monitoring, and data validation
• Developed scalable ML pipelines for time series forecasting, classification, recommendation systems, and predictive maintenance
• Integrated Explainable AI techniques such as SHAP and Grad-CAM; used frameworks like MLflow, Terraform, PyTorch
• Implemented real-world use cases such as passenger forecasting, hospital billing, medical image classification, and energy control

Results:

• Successful deployment of ML use cases in production
• Established reusable technical standards and components
• Increased efficiency through automation and improved forecasting models

Analysis of Tender Offers (Azure, GenAI)

Companies are regularly required to review complex and extensive tender documents (Tender Offers). Manual analysis is time-consuming, error-prone, and slows down decision-making. To address this challenge, a modular platform was to be developed that automatically processes documents, extracts relevant information, and provides AI-powered analyses. As Project Lead and Machine Learning Engineer, I was responsible for both the overall project management and the technical implementation. This included team coordination, defining the system architecture and roadmap, as well as designing and implementing the core ML and AI components.

• Led the project from requirements gathering to a production-ready prototype Designed and implemented a modular platform architecture using Python, FastAPI, Streamlit, and Docker
• Built pipelines for document classification, metadata extraction, and feature generation (e.g., keyword analysis, ranking algorithms)
• Integrated Generative AI (LLMs) to generate summaries of tender specifications
• Developed a multi-tenant RAG approach enabling chat-based interaction with documents
• Delivered REST APIs and an interactive Streamlit web application
• Integrated Azure Blob Storage and KeyVault for secure storage and IAM
• Established CI/CD pipelines with Jenkins, including automated unit & integration testing and documentation with Sphinx

Result:

• Delivered a scalable and secure platform for automated tender analysis
• Reduced manual effort through AI-driven extraction and summarization of relevant content
• Improved decision-making with structured preparation and visualization
• Created a sustainable technical foundation through modular architecture, multi-tenant capability, and cloud integration

Trustworthy AI in Agricultural Engineering (Research & AI Ethics)

In response to increasing regulatory demands, the agricultural sector required transparent, explainable, and trustworthy AI systems.

• Structured analysis of relevant use cases (e.g., classification of plant and animal conditions), partner acquisition and management
• Developed criteria for explainability, fairness, robustness, and auditability
• Technical coordination within the consortium and with external partners

Results:

• Project report served as a foundation for future AI certification approaches
• Visible contribution to AI compliance in public and industrial sectors

Additional Responsibilities at adesso SE

• Client acquisition & proposal development: Identification of potential client needs, technical consulting, and creation of tailored AI/ML solution proposals
• Internal LLM & MLOps community leadership: Strategic development of MLOps and LLM initiatives; knowledge building, best practices, coordination of internal MLOps projects
• AI.Lab mentoring: Recruiting, coordinating, and mentoring student developers on ML prototypes and innovation projects
• Generative AI showcases: Designing and presenting prototypes using Generative AI technologies (e.g., LLMs, LangChain, RAG) to demonstrate technical potential and foster innovation

As a Data Scientist, I was responsible for developing ML prototypes and implementing a data warehouse and data lake to support data-driven business models. I coordinated projects with external partners and developed ML approaches for GPS data analysis. Additionally, I played a key role in developing an IoT system and an Android app for work documentation.

• Developed multiple ML prototypes and a scalable data infrastructure
• Contributed to data strategy and project coordination with partners
• Created ML methods for activity classification based on GPS trajectories
• Designed and developed the backend of an IoT system
• Built an Android app for automated task documentation using Bluetooth beacons and GPS