I develop AI-driven industrial analytics solutions that bring together Large Language Models (LLMs), knowledge graphs, and unstructured data. My work spans prompt engineering, fine-tuning, and semantic vectorization techniques to transform complex operational logs and reports into actionable insights for asset optimization, risk reduction, and compliance. By integrating Generative AI outputs with ontologies and RDF data stores through SPARQL, I design Retrieval-Augmented Generation (RAG) pipelines that power intelligent query interfaces and decision support systems. With experience using RDFox, GraphDB, and AnzoGraphDB alongside modern LLM frameworks like GPT-4 and Mistral, I specialize in combining symbolic reasoning with generative AI to deliver explainable, business-ready solutions.

I worked on modeling and analyzing airline CO₂ emissions to explore how operational improvements could reduce carbon footprint. This involved generating and processing synthetic emissions datasets, building predictive machine learning models with scikit-learn and Pandas to forecast future trends, and simulating the effects of fuel-efficiency initiatives. To make the insights actionable, I designed a cloud-based analytics solution with Tableau dashboards that visualized temporal emission patterns and modeled scenarios, supporting data-driven decision-making in sustainability efforts.

I contributed to the administration and development of enterprise CRM for a global telecommunications client, building and maintaining SAP ABAP modules across both R/3 and S/4HANA environments. My work focused on creating modular, reusable components that streamlined ERP workflows and supported critical, high-availability business operations. Collaborating with cross-functional teams, I helped ensure system reliability, scalability, and alignment with evolving business requirements in a large-scale enterprise setting.

I worked on early healthcare AI initiatives applying deep learning techniques (Theano, TensorFlow, Keras) to medical records for disease prediction and pattern discovery. The project focused on combining natural language processing and handwriting recognition to process clinical documentation, while automating categorization in line with WHO’s ICD-10 standards. Using Salesforce to streamline storage and workflow integration, the goal was to reduce manual coding effort and improve accuracy in medical record management — an early step toward data-driven healthcare analytics.

I co-developed Robotic Process Automation (RPA) modules to streamline Accounts Receivable operations for FMCG clients. By automating payment record processing across vendor portals, the project reduced manual intervention and improved the speed and reliability of financial workflows in large-scale enterprise environments.