​- Quantified robustness and failure modes of control algorithms, informing requirements for sensor quality, timing, and system-level safety.

Developed control and planning algorithms for collaborative robot manipulators in automotive assembly, with a focus on compliant motion and safe human–robot interaction.

Applied model predictive control (MPC) and trajectory optimization to achieve robust, high-throughput assembly tasks under uncertainty.

​

Led architecture refactoring of core robot software to improve reliability, debuggability, and integration of advanced control algorithms into production systems.

Led diagnostic ML modeling for in-store activity using transactional and sensor-derived time-series data, focusing on causal analysis, anomaly detection, and forecasting to improve operations.

​

Designed and deployed production pipelines for activity monitoring and forecasting, collaborating closely with data engineering and product teams.

​

Matured MLOps practices across a distributed data science team, including standardization of model deployment, monitoring, and experiment tracking

Upgraded bank transaction parsing from rules-based features to embedding-based representations using NLP techniques, improving fidelity of customer banking history modeling.

​

Built and operationalized probabilistic graphical models for automated pricing optimization of credit lines and interest rates.

​

Mentored an intern in graph-based fraud detection using customer network segmentation and link analysis.

​

Delivered a European credit risk model for ING Bank via transfer learning from U.S. data, leading the full lifecycle from stakeholder alignment through deployment into production.

Designed a procedural system for dynamic simulated driving environments with virtual driver-in-the-loop control using Unreal Engine, supporting research in automated driving.