📌 Project Name
- doPILLganger
📋 Project Summary
- doPILLganger is an AI-powered mobile app. Its name combines “pill” with “doppelgänger”, reflecting its purpose – to help users find local medicines similar to the ones they usually take.
- The app was developed while I was living in Ireland, as part of the National AI Challenge hackathon organised by TechIreland.
- Users simply take a photo of their medication packaging, and the app employs image analysis and ingredient extraction to suggest local alternatives containing similar ingredients. It also displays nearby chemists and availability, helping people overcome language barriers and uncertainty when purchasing medicine abroad.
📅 Duration
- August 2024
🛠 Tech Stack
- GPT-4
- Prompt Engineering
- Vertex AI
- Google Cloud Run
- Java
- Spring Boot
👤 Role & Contributions
- Lived in Ireland and, despite being in an unfamiliar environment, proactively networked to connect Korean developers and designers. Recruited team members directly during the early stages for hackathon participation.
- Established collaboration channels to facilitate smooth communication. Coordinated team members with various roles, managed schedules, and led meetings to drive team operations.
- Worked as a backend engineer, primarily leveraging Vertex AI for image recognition and ingredient extraction, ensuring accurate processing of medication packaging photos.
- Designed and implemented APIs, developed core features, and managed the final deployment process.
🚀 Technical Highlights
- Built a serverless infrastructure using Google Cloud Run for rapid deployment and scalability.
- Utilised Vertex AI for image recognition and text extraction.
- Implemented medicinal ingredient refinement and medication matching logic using OpenAI GPT-4.
- Processed data in a structured JSON format to optimise integration with the frontend.
📈 Results
- Contributed to building a high-quality AI-based MVP within a limited timeframe.
- Achieved reduction in user anxiety, improved accessibility to information, and shortened medication search time.
- Focused on solving real-world problems based on use cases involving Korean international students and foreign travellers.
- Received positive interest and high evaluations in terms of practicality and technical excellence at the TechIreland-hosted hackathon.
🔗 Links
📸 Screenshots
