Lead Data Scientist @ Check24 Vergleichsportal

Full-time Munich, Germany AI Loan Comparison

• ML Systems & Business Impact
  • Implemented an MCP server using FastMCP, Pydantic AI, and LanceDB to inject company-specific context into AI coding tools, improving the relevance and accuracy of AI-assisted development.
  • Developed an NAO-based data analysis agent that lets non-technical users query and explore the company lakehouse in plain language, removing the dependency on data teams for ad-hoc insights.
  • Implemented a LightGBM model for ranking personalized loan offers. Boosted CTR by 1% and sales rate by 0.05%. Addressed challenges of highly dynamic offer sets and multi-stage ranking pipeline to scale to 50k+ offers.
  • Built a LightGBM model with quantile regression to predict customer-specific loan approval time, replacing heuristic approach and achieving 1 day MAE.
  • Designed and implemented improvements in model maintainability by automating training and data pipelines with MLflow, Airflow, and Kubeflow, resulting in faster training cycles, and reduced maintenance effort, allowing the team to focus more on experimentation and innovation.
• Technical Leadership & Ownership Impact
  • Owned the design and evolution of ML systems for personalized loan ranking and prediction of approval probability, interest rates, payout time, and required documents.
  • Led the design and development of a scalable data lakehouse platform using DBT, Trino, and Iceberg, standardizing data and feature pipelines across use cases and significantly reducing data preparation time.
  • Established MLOps best practices across the team, including model monitoring, A/B testing, and production observability using Prometheus and Grafana.
  • Drove adoption of model explainability, data quality, and drift monitoring (SHAP, Evidently), improving transparency and reliability of deployed systems.
  • Introduced engineering standards and tooling (CI/CD improvements, SonarQube, pre-commit hooks, testing practices), improving maintainability and consistency across projects.
• Team & Organizational Impact
  • Acted as the data science representative in cross-team technical discussions, influencing architectural decisions and aligning priorities across teams.
  • Led and mentored a team of 3 data scientists and 2 engineers, supporting hiring, onboarding, and career development. Promoted a culture of experimentation and continuous learning.

Senior Data Scientist / Machine Learning Engineer @ e-bot7

Full-time Munich, Germany AI Chat Bots

• ML Systems & Business Impact
  • Fine-tuned and trained transformer-based models (BERT) for semantic similarity and question answering, improving sentence similarity performance by 20 percentage points.
  • Introduced multilingual embedding models (DE/EN/FR), improving performance (+3%) while reducing GPU costs.
  • Reduced embedding dimensionality by 74% using knowledge distillation, significantly lowering storage and inference costs.
• Technical Leadership & Ownership Impact
  • Drove key infrastructure decisions around model training and inference, including adoption of Kubeflow based pipelines and AWS services (Glue, Athena, Lambda) and scalable serving with Triton Server and gRPC.
  • Owned the design and evolution of semantic search and question answering RAG systems, driving improvements in both model performance and system architecture.
  • Led the transition of embedding management from FAISS to Elasticsearch, improving system robustness and operational efficiency.
  • Led the transition from legacy systems to a modular, event-driven architecture, improving scalability, maintainability, and system flexibility.
• Team & Organizational Impact
  • Contributed to hiring and scaling the engineering team, including automating parts of the hiring pipeline and interviewing candidates.
  • Served as Python Chapter Lead, mentoring engineers through regular 1:1s and supporting career development and knowledge sharing.

Software Engineer II @ Lybrate

Full-time New Delhi, India Online Doctor Appointment

• Developed web applications using Java Spring, Hibernate, MySQL, MongoDB, Apache Solr. Boosted user engagement by implementing a personalized facebook like feed of health tips and QnAs.

Software Engineer I @ Snapdeal

Full-time New Delhi, India E-commerce

• Built scalable and highly available backend systems with Java, Spring, Hibernate, and MongoDB hosted on AWS.

Master of Science in Informatics

Technical University of Munich Munich, Germany Grade: 1.4, Passed with Distinction

Relevant Courses – Machine Learning, Deep Learning in Computer Vision, 3D Scanning and Motion Capture, Variational Methods, Medical Augmented Reality, Cloud Computing.

Master Thesis – Aspect-based Sentiment Detection using Deep Neural Networks and Transfer Learning (2019). A novel approach to jointly detect aspect and the corresponding sentiment in a given sentence using DNN and transfer learning.

Bachelor of Technology in Software Engineering

Delhi Technological University New Delhi, India Grade: 75.9%, First Class Distinction

Relevant Courses – Obejct Oriented Programming, Operating Systems, Databases, Engineering Mathematics.