DDI-Checker — MCP Server (Prototype) for Interaction Research
Prototype of an MCP server for researching drug-drug interactions in public US data (openFDA, FAERS).
Prototype of an MCP server for researching drug-drug interactions in public US data (openFDA, FAERS).
End-to-end development of regulatory AI workflows: MedDRA mappings in minutes instead of half an hour, an automated compliance audit, EMA SPOR integrations.
Development of a novel pharmaceutical dosage form with a validation concept to pharmacopoeia standards.
International market analysis and validation concept for automated filling systems in pharmaceutical compounding.
Binary classification of whether a question is a health topic — with a deliberately small model: a frozen DistilBERT (66M parameters), with only a 769-parameter head being trained. ~98% accuracy.