Jens D. - Product Owner & Senior Data Scientist

Frankfurt, Germany

Experience

Jun 2023 - Present

2 years 7 months

Frankfurt, Germany

Product Owner & Senior Data Scientist

Legal Tech

Led an international team of six developers in a Scrum environment
Defined the project’s strategic goals in coordination with stakeholders and the development team
Performed prompt engineering on language models to improve answer accuracy and relevance
Implemented LangChain components for a RAG chatbot to answer legal questions
Technologies: GPT-4, LangChain, Python (Pandas, sklearn, streamlit), Docker, GitLab, ChromaDB

May 2023 - Apr 2024

1 year

Frankfurt, Germany

Senior Data Scientist

Zentralbank

Developed a scalable, high-performance enterprise search solution
Implemented a Retrieval-Augmented Generation (RAG) model to deliver valid, context-aware answers to document queries
Built and managed messaging queues to ensure reliable, scalable data processing and transfer between system components
Created RESTful APIs to provide search functionality and integrate the enterprise search solution into existing applications and systems, including security and authentication
Technologies: ElasticSearch, Kibana, LLaMA, SQL, FastAPI, Docker, Python (Pandas, sklearn, PyTorch), HuggingFace

Sep 2022 - Apr 2023

8 months

Senior Data Scientist

Finanzaufsichtsbehörde

Developed and deployed an early warning system using structured and unstructured data to monitor fund default risk
Built a GPT-4 based chatbot for the regulatory authority to answer questions on annual and quarterly reports
Implemented and configured automated CI/CD pipelines to automate build, test, and deploy processes
Collaborated closely with subject matter experts to understand requirements for the early warning system
Technologies: GPT-4, LangChain, Python (Pandas, NumPy, PyTorch, sklearn), SQL, GitLab, Docker, Kubernetes, Apache Spark, ChromaDB

Dec 2021 - Aug 2022

9 months

Berlin, Germany

Senior Data Scientist

Öffentliche Behörde

Led the project, held regular client meetings, and ensured all requirements and expectations were met
Developed and trained models to analyze economic and financial market reports
Improved model performance through hyperparameter tuning, feature engineering, and regularization
Worked with experts to validate results and adapt models to the agency’s specific needs
Technologies: Python (Pandas, NumPy, SpaCy, sklearn, Keras), HuggingFace, GitLab, Docker

Mar 2021 - Nov 2021

9 months

Salzburg, Austria

Senior Data Scientist

Getränkehersteller

Analyzed historical sales data to identify patterns, trends, and seasonal effects impacting sales
Applied time series techniques like ARIMA and Exponential Smoothing, plus advanced ML models like Random Forests and LSTM to improve forecast accuracy
Integrated external data (weather, marketing campaigns) to further enhance sales predictions
Conducted sensitivity analyses and scenario modeling to spot risks and opportunities early and develop strategies
Technologies: Python (Pandas, NumPy, seaborn, sklearn), GCP (Dataproc, BigQuery, Cloud Functions, Vertex AI), SQL, GitLab

May 2020 - Feb 2021

10 months

Zürich, Switzerland

Senior Machine Learning Engineer

Universalbank

Built and implemented preprocessing pipelines to standardize and structure traders’ communication data
Used the pre-trained FinBERT model to generate word embeddings from finance texts
Developed models for network analysis, anomaly detection, and clustering
Created and automated end-to-end workflows for model training, validation, and deployment
Technologies: Python (SpaCy, sklearn, TensorFlow), Hugging Face, SQL, ElasticSearch, Docker, Kubernetes, GitHub, Jenkins, MLflow

Oct 2019 - Apr 2020

7 months

Utrecht, Netherlands

Data Scientist

Behörde für Verkehrsdaten

Designed and implemented cloud-based system architectures on Azure
Set up and configured Kubeflow and MLflow to manage and automate ML workflows
Built and trained ML models to detect unusual traffic patterns and events
Developed and ran tests to ensure solution functionality, reliability, and security
Technologies: Python (TensorFlow, PyTorch, Pandas, NumPy), Azure (Kubernetes Service, DevOps, Storage), Kubeflow, MLflow, Helm

Mar 2019 - Sep 2019

7 months

Stuttgart, Germany

Machine Learning Engineer

Automobilkonzern

Collected and cleaned historical sales data and external factors like market trends, economic data, and seasonal influences
Identified and engineered relevant features to boost model accuracy
Developed and trained various ML models for sales forecasting, including specialized models like Prophet and ARIMA
Implemented an Explainable AI module using SHAP to increase transparency and interpretability of results
Technologies: Python (Prophet, statsmodels, Keras, Pandas, NumPy, SHAP), SQL, GitLab

Aug 2018 - Feb 2019

7 months

Düsseldorf, Germany

Data Scientist

Asset Manager

Developed containerized microservices, including APIs and test specs, for named entity recognition using custom and pre-trained AI models
Built models to score fund report sentiment
Identified and engineered key features from text data to improve model performance and selected top features for training
Optimized hyperparameters through systematic search and advanced methods like Bayesian optimization
Technologies: Python (Pandas, NumPy, NLTK, SpaCy, TensorFlow), Flask, Azure (Databricks, Cognitive Services, Machine Learning, DevOps)

Jan 2018 - Jul 2018

7 months

Vienna, Austria

Data Scientist

Universalbank

Designed a document data processing workflow with seamless OCR and NLP module integration
Implemented OCR algorithms to auto-recognize text in image files (tif, jpg, png), containerizing OCR microservices with Docker
Built NLP models to extract information from recognized text
Extracted key features from OCR and NLP data to boost model performance and improve information extraction
Technologies: Python (Tesseract, SpaCy, NLTK, Pandas, NumPy), Docker, Kubernetes, GitLab

Jun 2017 - Dec 2017

7 months

Berlin, Germany

Data Analyst

Kreditbank

Developed and validated credit risk models in Python, including Monte Carlo simulations for various risk scenarios
Used SQL to manage and query large datasets, then prepared, cleaned, and explored data in Python to spot key features and patterns
Validated models via backtesting and historical data analysis, then fine-tuned based on validation results
Integrated models into the bank’s IT system for production use and set up continuous monitoring and optimization
Technologies: Microsoft SQL, Python (Pandas, NumPy, SciPy, sklearn, Seaborn), GitLab, Docker