Provide advanced machine learning solutions for Stadtwerk am See:• Constructed a scalable Q&A system using Haystack, LangChain, and Streamlit, reducing user queryresponse time by 25%. Leveraged Mistral for knowledge base and information retrieval, deployed onHugging Face Spaces.• Enhanced answer retrieval accuracy by 20% by fine-tuning transformers (e.g., BART) within Haystackand integrating Retrieval-Augmented Generation (RAG) techniques.• Achieved 90% precision in the Zero-Shot Post Classification System using RoBERTa XLM withoutlabeled training data, demonstrating the effectiveness of large language models (LLM).Deliver consultancy to one of the largest DIY retailers in Europe (OBI Germany):• Designed and implemented a data pipeline using AWS Kinesis for real-time ingestion of 500,000+daily sales and customer records from POS systems, and AWS S3 for scalable storage.• Developed ETL workflows with AWS Glue and AWS Lambda (Python) to transform and aggregatesales data, reducing processing time by 40%. Integrated 1 TB of monthly data into AWS Redshiftfor advanced analytics.• Utilized Apache Spark within AWS Glue for large-scale data processing and complex transformations,enhancing data accuracy and processing efficiency.• Created interactive dashboards using AWS QuickSight to analyze sales trends, customer behavior,and inventory performance, leading to a 15% improvement in sales efficiency.

Analysis of Job traces in HPC Systems