Curriculum Vitae
Senior Data Scientist and Machine Learning Engineer with a Ph.D. in Statistics specializing in strategy implementation, ML solution development, data processing, statistical modeling, data-driven decision-making, strategic consulting, and automation. Many years of experience in developing and improving data management processes, with a focus on improving business outcomes in the financial and insurance sectors, and a strong background in project management and stakeholder communication.
Experience
Senior Data Scientist
Churn & Customer-Lifetime-Value | AZ Ireland
- - Developed a churn strategy to retain customers and increase annual premiums based on customer lifetime value.
- - Led technical team implementing churn use case in the context of outbound call center.
- - Conducted the implementation of A/B testing to prove the impact of the models compared to business-as-usual.
- - Oversight and testing of implementation and E2E integration on ADP’s Azure ML Studio and Azure OpenAI.
- - Presented to a variety of audiences, including senior management, department heads, and cross-functional teams, on the strategy, results and insights of the model.
Senior Data Scientist
GenAI Driven Reason-To-Believe (GenRTB) | GSMD & AZ Direct Italy
- - Developed a dashboard combining customer information, 3rd party data, and AI model outputs with LLMs.
- - Provided understandable insights, arguments, and facts about potential risks.
- - Demonstrated the value of GenRTBs in the H4 & H5 community.
AI Driven Customer Advisory (Lizzy) | GSMD & ABV
- - Developed backend solution using Azure OpenAI services, i.e., Whisper, Dalle-3, GPT4.
- - Utilizing predictive machine learning models with Azure OpenAI services.
- - Leveraging Retrieval Augmented Generation (RAG) techniques to avoid hallucinations of the GPT models.
- - Developed frontend application and ensured a seamless connection to the backend.
- - Deployment of the solution (backend & frontend) on GDP.
Global Churn & Cross-Sell Models | AZ Italy, AZ Spain, AZ France & AKM
- - Designed and implemented global churn and cross-sell predictive models for use across multiple OEs.
- - Initiated the collaboration with the DS communities of the big four OEs to set the best standards for the global models.
- - Packaging of global churn and cross-sell models for easy integration into multiple OE IT environments.
Churn & win-back | AZ Direct Italy
- - Developed and operationalized omnichannel predictive models for customer churn and re-engagement, resulting in a 9% increase in customer retention. Collaborated with cross-functional teams to implement data-driven interventions across digital, call center, and marketing channels.
- - Presented model findings and insights to a wide range of audiences, including senior management, department leaders, and interdisciplinary teams.
- - Enhanced data quality and reporting accuracy by developing advanced data handling processes.
- - Collaborated closely with IT and business departments to streamline data workflows, improving the accuracy and efficiency of quarterly reporting processes.
Lead Rating | AKM & AZ France
- - Leveraged new AI-driven approaches to calculate the conversion probability of incoming new leads and allocate leads to the best-fit agents.
- - Designed and developed predictive models for Lead Rating.
- - Built and deployed the model APIs on AWS S3 as part of the project infrastructure.
- - Developed the global Lead Rating model that can be seamlessly integrated into other Operating Entities as part of the AI in Box initiative.
- - Containerized the entire Lead Rating process using Docker for easy deployment and scalability
- - Presented model findings and insights to various audiences, including senior management, department heads, and cross-functional teams.
Research Associate
Ph.D. candidate, dissertation topic: Modeling Unobserved Heterogeneity in Social Network Data Analysis | Lecturer for introductory and programming courses | Lecturer for classes at the Depart- ment of Statistics | Supervisor of bachelors and masters theses | Collaboration on the monitoring report 2019 of the Federal Environment Agency (Umweltbundesamt) | Statistical consulting of > 50 internal small and large projects | Preparation of scientific publications.
Student Assistant StaBLab
Statistical consulting and data analysis for many external and LMU internal projects.
Managing Director
Company in retail and gastronomy. Locations in Munich, Neubiberg and Baldham.
Certifications
Certificate
Certified Machine Learning Engineer: 💻 Bash & Linux | 🚀 FastAPI | 🌐 Flask |🔒 API Security | 🐳 Docker |🚦 Jenkins | ☸️ Kubernetes | 🌬️ Airflow | 📊 MLFlow | 🚀 Streamlit | 📈 Dash
Education
Ph.D.
Degree: Dr. rer. nat., Thesis topic: Modeling Unobserved Heterogeneity in Social Network Data Analysis.
Master’s Degree in Biostatistics
Degree: M. Sc., Biostatistics, Thesis: Separable Temporal Exponential Random Graph Models.
Diploma in Psychology
Degree: Diplom in Psychology, Thesis: Conflicts in Organizations and The Ways of Overcoming Them.
Awards
Best Technical Blueprint
Allianz Group GenAI Hackathon 2023: Winner in the category BEST Technical Blueprint among 22 participating teams.
Best Poster Award
Iterative Estimation Approach for Exponential Random Graph Models (ERGMs) with Nodal Random Effects. International COSTNET Conference 2018, September 26-28, in Warsaw Poland.
Communication
Conferences
- Oral presentation at the DAGStat Conference 2019 in Munich, Germany
- Oral presentation at the summer retreat 2019 of the department of statistics in Holzhausen, Germany
- Oral Presentation at the COSTNET Conference 2019 in Bilbao, Spain
Posters
- Poster at the COSTNET Conference 2018 in Warsaw, Poland
- Poster at the MCML: Munich Center of Machine Learning 2019 in Munich, Germany
Publications
Fritz C., De Nicola G., Kevork, S., Harhoff D. and Kauermann G. (2023).Modelling the large and dynamically growing bipartite network of German patents and inventors. Journal of the Royal Statistical Society. Series A: Statistics in Society.
Kevork, S. and Kauermann, G. (2022). Bipartite Exponential Random Graph Models with Nodal Random Effects. Social Networks, 70:90-99, ISSN 0378-8733, https://doi.org/10.1016/j.socnet.2021.11.002.
Kevork, S. and Kauermann, G. (2021). Iterative Estimation of Mixed Exponential Random Graph Models with Nodal Random Effects. Network Science 9, 478–498. https://doi.org/10.1017/nws.2021.22.
Witting, M., Filimon, S., Kevork, S. (2021). Carry along or not? Decision-making on carrying standard avalanche safety gear among ski tourers in a german touring region. Safety Science, 105406, ISSN 0925-7535, https://doi.org/10.1016/j.scci.2021.105406.
Pöllath, N., García-González, R., Kevork, S., Mutze, U., Zimmermann, M. I., Özbasaran, M., Peters, J. (2021). A non-linear prediction model for ageing foetal and neonatal sheep reveals basic issues in early neolithic husbandry. Journal of Archaeological Science, 105344, ISSN 0305-4403, https://doi.org/10.1016/j.jas.2021.105344.
Freick, M., Weber, J., Passarge, O., & Kevork, S. D. (2017). Effects of erythromycin on milk yield, rumen contraction rate and blood beta-hydroxybutyrate concentration in dairy cows after laparoscopic reposition of the left displaced abomasum. Veterinarni Medicina, 61(8), 413-420.
Languages
Interests
Theater Enthusiast: I find immense joy in the captivating world of theater, where I can immerse myself in new and exciting narratives. Passionate Traveler and Culture Explorer: Traveling and immersing myself in diverse cultures is an integral part of my quality of life. With my proficiency in multiple languages, I relish the opportunity to connect with people from different backgrounds, effortlessly bridging cultural divides. Movie Aficionado: Whether it’s cozy movie nights with my family or exploring the latest cinematic masterpieces, I have a genuine appreciation for the art of storytelling through film.
Motto
"If you can't explain it to a six year old, you don't understand it yourself"
Albert Einstein