Spearheaded the modernization of the ITER fusion community's tech stack by architecting and deploying its first-ever Python-based backend workflows, establishing a new standard adopted by international researchers. Engineered highly scalable data pipelines for the ingestion and processing of massive, complex fusion reactor datasets, bridging legacy transport codes with modern Python frameworks. Architected a robust data aggregation system to ingest multi-code simulation results into a centralized database, directly enabling large-scale model validation and future machine learning applications.

Designed and implemented a Redis-based backend caching system with selective invalidation, drastically reducing latency for real-time rule triggering. Redesigned tree-based data structures to enable parallel node computation, accelerating health metric calculations by over 250% for 1,000+ users. Integrated the rule-evaluation engine with mobile app microservices (notifications, emails, UI pathways) via RESTful APIs and MCP servers. Developed comprehensive health profiling and reporting dashboards aggregating historical user data across organizations, extracting trends and insights with LLM APIs (OpenAI, Anthropic). Integrated third-party health data APIs, including Apple Health, Google Fit, and clinical lab results.

Researched, developed/fine-tuned, and deployed diverse ML/AI models (BERT, LSTM, Random Forest, Gradient Boosting, LLMs) tailored to client needs across health, chemical, and administrative domains. Co-developed an automated ML/AI workflow Python package for model training/fine-tuning, evaluation, deployment, and monitoring across cloud platforms (GCP, VertexAI, BigQuery, S3) or own infrastructure. Engineered a secure, multi-lingual NLP pipeline using Tesseract OCR and BERT to extract unstructured patient records into a local PostgreSQL database, ensuring strict data privacy and reducing manual data entry time for oncology staff by 80%. Created an early-stage conversational interface (LLM) to democratize AI project automation for non-technical users, forming the basis for a virtual AI engineer.

Ph.D. with Max Planck Institute for Plasma Physics (via Technical University of Munich). Research topic: Energetic particles stability in a magnetic confinement device (ITER). Architected the first-ever Python-based backend workflows for the ITER fusion community and engineered scalable data pipelines to bridge legacy transport codes with modern Python frameworks.

Development of Integrated Modeling workflow to describe fast ion stability in ITER using IMAS framework. Linear and non-linear Energetic Particle Physics in various ITER Scenarios. Thesis: Workflow-based energetic particle stability analysis of projected ITER plasmas

The group performs theoretical calculations and computational simulations including method development on a variety of molecular systems under the supervision of Prof. Dr. Ulrich Kleinekathofer. Part of the team which investigates ”Transport of ions and molecules across nanopores”. Molecular Dynamics and Brownian Dynamics simulations. 3D visualization of molecules and their properties

For Applied Mathematics and Mathematical Concepts in Sciences Courses: Wrote the support slides of the course. Helped preparing the assignments for the course. Teaching support for the course including homework grading, exams grading and tutorials.