My Work
Featured Projects & Case Studies
Explore how I apply data strategy, engineering, and AI to solve complex business challenges.
Predictive Retention MVP for a Leading E-commerce Platform
An independent project I developed to solve customer retention challenges for About You.
The Challenge
E-commerce platforms face significant challenges in retaining customers. Proactively identifying users at high risk of churn is critical for targeted marketing and improving customer lifetime value.
My Solution
I architected and built a full-stack MVP, including a Python-based machine learning model (XGBoost) to predict churn probability, a FastAPI REST API to serve predictions. Both useful to create a visualization tool to empower teams to make better marketing investment decisions and also ready to use a macroservice for a personalized user experience. The API was fully documented and tested for production-readiness.
Impact & Results
My model demonstrated a strong predictive power, capable of identifying at-risk customers with over 85% accuracy (AUC). This validated the feasibility of a data-driven retention strategy, enabling a potential 15-20% reduction in untargeted marketing spend.
Key Technologies
Python (Scikit-learn, Pandas, XGBoost), FastAPI, Docker.
AI-Powered Chatbot for Public Transport Regulations
A self-initiated project to demonstrate a GenAI solution for Berlin's BVG.
The Challenge
Berlin's public transport rules are extensive and can be confusing. Finding a quick, accurate answer often requires searching through complex PDF documents, leading to a poor customer experience.
My Solution
I built a proof-of-concept using a RAG (Retrieval-Augmented Generation) architecture. The system ingests official BVG rule documents into a FAISS vector database. A Flask API exposes the backend, allowing a Streamlit-based web interface to ask questions in natural language.
Impact & Results
The prototype successfully answered complex queries like 'Can I take my bike on the S-Bahn at 8 am on a weekday?', providing instant, cited answers. This demonstrated a clear path to significantly improving user experience and reducing the burden on support teams.
Key Technologies
Python (LangChain, Transformers), FAISS (Vector DB), Flask, Streamlit.
Interested in my work?
Let's connect and discuss how my skills and experience can bring value to your team.