Replacing legacy systems and shadow IT with one central cockpit for steering the operational processes of key-account heating plants

• Owned product strategy and roadmap in an agile Scrum setup; led a team of 3 from use-case discovery to delivery
• Mapped data across a fragmented landscape of legacy systems and shadow IT and consolidated it into one integrated data model
• Delivered the cockpit for monitoring and steering heating-plant processes, replacing legacy tools and shadow-IT spreadsheets
• Prioritised use cases with business units, broke them into work packages and aligned stakeholders up to management

Implementation and stabilisation of a new securities settlement system, combined with data-driven process optimisation to raise operational efficiency

• Supported the implementation and stabilisation of the new settlement system
• Analysed settlement processes and data flows to identify bottlenecks and optimisation use cases
• Designed and implemented use cases that measurably increased operational efficiency

Quantitative score and decision framework to allocate funding where it achieves the highest biodiversity impact per euro

• Developed the impact score based on literature research and expert interviews
• Built a decision framework rating intervention types and allocation options; validated it in a proof of concept on real projects
• Led requirement engineering with the client and documented the framework for stakeholder use

• Consolidated heterogeneous production and complaint data, incl. data-quality assessment
• Applied and refined LLMs to auto-categorise complaints at scale as input for downstream models
• Designed and backtested anomaly and scam detection models (KPIs, indicators, rules) with subject-matter experts
• Delivered insights via Power BI dashboards and executive presentations

• Built the modelling framework and ran data QA for the performing and in-default LGD models
• Computed realised LGDs incl. expected recoveries on open defaults; estimated sub-models via uni- and multivariate analysis
• Defined calibration segments and calibrated all models, incl. in-default models per reference date; delivered regulatory documentation

• Built a behavioural PD model (logistic regression) plus ML challenger models steering company-wide collections decisions
• Set up a behavioural data warehouse, derived risk factors and built a Python dashboard for risk-factor analysis
• Designed collections KPIs, monitoring and automated SAS reporting across operations and risk
• Evaluated A/B tests to optimise collections strategies; presented results up to upper management