Throughout 2024, I worked on several applied AI projects combining natural language understanding, large language models, and generative image systems. These projects explored different interaction models and problem domains.
1. Enterprise QnA Chatbots for Internal Knowledge Access
Development of conversational bots for querying internal company databases using natural language. Users could retrieve structured information such as names, descriptions, or dates using flexible question formats.
- Technologies: Microsoft Bot Framework Composer, LUIS (Language Understanding), Azure Bot Service, static tabular data
- Channels: Microsoft Teams, Telegram
- My Role: Chatbot design, language model tuning, Azure integration
2. Medical Image-Enhanced LLM Evaluation
Created a prototype interface allowing users to upload medical images and receive structured feedback from an OpenAI multimodal model. Responses were formatted into user-focused tables for clarity and actionable insight.
- Technologies: OpenAI GPT Vision
- My Role: Prompt engineering, interaction logic, results design
- Use Case: Exploratory tool for image-based medical feedback (non-diagnostic)
3. Generative Catalog Builder with DALL·E
Built an interactive pipeline to generate visual + descriptive entries for custom catalogs. GPT was used to generate item descriptions and context, while DALL·E rendered matching visuals — enabling creative studios and researchers to ideate faster.
- Technologies: OpenAI GPT-4 + DALL·E
- My Role: Prompt design, content logic, generative output flow
- Use Case: Tool for creative professionals and early concept visualization