● Own architecture, hands-on implementation, deployment and production support for AI products while leading two developers. ● Built an AI quality platform that combines CRM data and call transcripts to identify missing or conflicting facts and produce actionable reports for leadership. ● Designed a stateful multi-stage LLM pipeline with evidence, confidence scoring, human review, idempotency, retry/fallback and independent recovery; it processes hundreds of cases monthly. ● Created a fact-level evaluation harness using 60 manually labelled reports; extraction quality reached up to 0.93 F1.

I worked as the lead developer in a four-person team. We developed a full-stack web application for a microelectronics testing framework. We also integrated an AI assistant that could answer users’ questions and interact with external services. As a side project, I developed a custom AI coding assistant for VS Code to help engineers get started with our internal framework. It explained available methods, suggested suitable APIs, and provided code completion.

● Built Python 3D-generation workflows and C#/.NET integrations connecting Revit data with Unity visualisation.

● Designed, tested and debugged C# and Python/pyRevit automation plugins that reduced MEP modelling time by 15-40%.