At Synthetica, I build production LLM and agent systems for enterprise clients, from custom orchestration engines to framework-based agentic RAG. I co-architected a framework-free engine that routes work across specialized sub-agents and preserves state across stateless requests.
At Mobile Technology SA, I designed and maintained a bilingual Greek-English RAG system processing more than 20,000 documents. I also built guarded text-to-SQL, incremental document ingestion, and a custom asynchronous FastAPI backend.
Across these client engagements, I co-built a reusable document-AI platform with configurable stages for segmentation, extraction, generation, and validation. I also developed LLM evaluation and regression-testing tools to assess retrieval precision, faithfulness, and instruction-following.
Before focusing on AI engineering, I worked in data science at AETHON Engineering and built the backend and data layer for TransiTool. My background combines data science with civil and transportation engineering, including thesis research on lane-change prediction using LSTM networks.
