Frankfurt Rhine-Main Metropolitan Area
Subscription Service Management System (Java Backend) • Developed features in a production Java backend, deepening expertise in system architecture and API design within a microservices environment • Optimized service and database query performance (PostgreSQL index tuning) and debugged issues in complex, distributed systems • Applied a multi-level testing strategy (unit, integration, and acceptance tests with BDD/Gherkin) to ensure service quality Proof of Concept: Internal RAG System with Open-Source LLMs • Designed and implemented a retrieval pipeline using LlamaIndex, vector database, and reranker for enterprise-internal knowledge bases • Local LLM inference with quantized open-source models and performance optimization via GPU offloading • Leveraged built-in evaluation metrics (faithfulness, relevancy) and Phoenix for automated quality assessment and observability, with prompt engineering including guardrails • FastAPI backend with CI/CD pipeline and automated code-quality checks
Data Order Management System • Implemented REST API services using Azure technologies (Functions, Service Bus) and migrated a synchronous workflow to an asynchronous architecture • Built an end-to-end observability platform with Application Insights and automated alerts • Defined and implemented a release strategy and multi-stage quality gates in CI/CD pipelines, supplemented by introducing tools such as ShellCheck, Vulture, and Dependabot to improve code and dependency quality
• Develop, implement, deploy, and test solutions for data-driven decision support in competitive and market analysis • Data acquisition from external sources using automated methods such as document and web scraping with Python • Data preprocessing, harmonization, and storage in a MySQL database according to a co-designed data model • Analysis of data in Python using various methods from simple statistics to Natural Language Processing • Visualization of results in Tableau dashboards
• Construct Attention-based Encoder-Decoder Network with a Bash script, Kaldi, and Pytorch to enhance and recognize speech • Construct a transformer network to improve the recognition results of enhanced speech
• Development of a script with Python for translating unstructured PDF into structured text data • Analysis of images with OpenCV to improve errors identified by the OCR tool Tesseract • Implementation of TensorFlow to generate Word embedding to generate similarity of paragraphs • Implementation of Web App with Flask, HTML, CSS, Bootstrap and JavaScript • Deployment Web App with Azure Linux service, Gunicorn and Apache2
• Development of a script with Python for data translation of electronic components generated by test benches • Definition of the conversion process to execute multiple parser scripts of different electronic components • Function implementation of a LabVIEW tool for generating XML data under a standard structure • Adaptation of the RapidMiner process for monitoring the data status • Adaptation of the component data status (Tableau Dashboard)