Agentic MLOps

To make demand forecasting more accessible, I developed an agent-based layer that simplifies the complexity of MLOps for Demand Forecasting domain. By implementing the Model Context Protocol (MCP) for interfacing APIs and tools and Agent-to-Agent (A2A) protocol for agent orchestration, I built a system where specialized subagents collaborate to handle from configuration to data analysis. Using Retrieval-Augmented Generation (RAG), the system surfaces in-house best practices at the moment of decision-making, ensuring that complex model adjustments are reproducible, deterministic, and transparent for users.

Sustainable supply chains

At my current position with Blue Yonder I support retailers and grocers from the replenishment and demand forecasting side up to intelligent pricing solutions. Incorporating the solutions that use modern technologies such as cloud infrastructure and machine learning help companies save costs and compete within a world with many uncertainty factors to manage. It also helps us reduce waste and bad business practices that hurts our environment and lives.

Hunting for Higgs

During my time as a physicist I contributed to the search and observation of the Higgs particle. This mysterious particle of nature helps us understand why atoms and their subparts have mass and will be a gateway to new technologies as we learn more about it. To study this particle I learned a lot about experiment design, complex data in large volumes and how to analyse them using deep learning methods.

Building virtual skies

As an entrepreneur I founded a company that specialized in creating full dome projection experiments. The devices and software I helped write and design are used in small and large planetarium projection systems created by the founded company.