*) MLOps Platform: Building end-2-end MLOps workflows/pipelines for the containerized deployment of ML applications backed by GPU compute TechStack: {Kubernetes, Kustomize, Gitlab CI/CD, FastAPI, NVIDIA Triton Inference Server, Grafana}*) Intelligent Document Processing Monitoring: Implementation of monitoring dashboards for the in-house IDP system TechStack: {Plotly Dash, Javascript}*) Intelligent Document Processing System: Setup of a scalable in-house IDP system, processing over 3 million documents yearly across major document classes since 2021 TechStack: {Python, ORACLE, celery, RabbitMQ, Flask}*) OCR Engine Development: Implementation of two OCR Engine Services, Tesseract and ABBYY Finereader Engine 12 TechStack: {Python, Java}*) Intelligent Document Processing Research: Development of deep learning models for IDP services, i.e., document classification, document extraction, and document separation/structuring TechStack: {Python, PyTorch, TensorFlow, Transformers Hugging Face}

*) Intelligent Document Processing For A Large German Bank: Setup of an IDP system for trade finance documents, tree-based ensemble learning on top of deep learning microservices TechStack: {Python, Docker Compose, Gitlab CI/CD, PyTorch, TensorFlow, spaCy, Flask}*) Liquidity Forecast For A Large German Electric Utility Company: Developing ETL processes, building regression models for forecasting consumption and price data, modeling of balance sheet / P&L / cash flow positions, setup of a customer dashboard TechStack: {Python, Azure VM, Snowflake, scikit-learn, Power BI}*) NLP Prototyping: Building several NLP-based applications TechStack: {Docker Compose, Django REST framework, Elasticsearch, R-Shiny, React}

Abstractive Summarization: Extension of state-of-the-art Deep Learning Sequence-to-Sequence neural networks to process text across content windows, such that inference on arbitrary long texts becomes feasible:- https://www.inovex.de/blog/text-summarization-seq2seq-neural-networks/- https://arxiv.org/abs/2004.03324