- Developed an LLM agent that responds based on the company's internal database- Classified auction histories using image embeddings and metadata- Identified similar artworks using image embeddings

- Trained VQ-VAE for speech tokenization- Implemented a speech synthesis model- Improved text and speech alignment for enhanced TTS (Text-to-Speech) service- Generated new voices by isolating emotions from one voice and synthesizing it with another

- Designed a service to understand business processes using deep learning models- Designed a real-time data processing service