At BonZuttner Inc., I engineered backend components for a LangGraph-based agentic RAG system, including site-isolated content ingestion and structured content processing. I worked on hybrid retrieval using dense embeddings, BM25, Reciprocal Rank Fusion, query-aware boosting, and Weaviate.
I also developed Python REST APIs and backend logic for IoT systems, and independently executed an end-to-end migration of a legacy multi-tenant pharmaceutical platform to a local PostgreSQL stack. My work there included REST APIs and RBAC workflows, as well as backend modules and automated tests for an AI-powered recruitment platform.
On the AI-Powered Adaptive Examination Platform graduation project, I led technical planning for a five-person team, designed the backend architecture, and developed an adaptive assessment engine using IRT/CAT. I also developed an AI question-generation workflow with retrieval, reranking, evidence validation, difficulty calibration, repair, and duplicate detection.
