Overview
TercüMed AI is an AI-supported healthcare product that helps medical institutions run multilingual medical document translation faster, more securely, and at corporate quality. The product focuses on sensitive health content like epicrisis reports, lab results, consent forms, discharge summaries and patient information documents.
Problem
When medical translation runs manually inside a healthcare institution, you get time loss, terminology errors, format breakage, data security risk and operational bottlenecks. General-purpose translation tools fall short for health data, medical terminology, patient privacy and the integration needs of a hospital.
System
I shaped the product around document upload, language selection, translation output, terminology control, format preservation, human review, export and corporate logging. I pulled HBYS/API integration, KVKK/GDPR compliance, role-based access, audit trail, data security and medical terminology accuracy into the product scope.
For local LLM evaluation I built a benchmark logic for how Llama, Mistral, Qwen, Gemma and Phi families could be compared for this product specifically. I worked through translation quality, terminology accuracy, clinical error risk, format preservation, data security and cost as product-level metrics, and took a central role in the product's modeling decisions.
Outcome
TercüMed AI is positioned not as a general-purpose translation tool but as an AI-supported medical translation platform that meets the security, quality, integration and auditability needs of healthcare institutions. The product narrative was aligned with the language used by technical teams, healthcare institutions, and decision-makers.
My role
Product scope, technical roadmap, software architecture, use cases, health data requirements, KVKK/GDPR and HBYS integration expectations, LLM benchmark approach, technical evaluation reports, data collection protocol, consent form, KOSGEB / R&D documentation, and stakeholder presentation outputs.