Automated Semantic Search and Matching Engine (FAISS + SBERT)
Built a semantic product recommendation pipeline using FAISS for vector indexing and Sentence-BERT for generating dense embeddings. The system matches user queries to relevant products based on meaning not just keyword overlap, enabling highly relevant recommendations at scale. Fully deployed and live, demonstrating real-world vector database architecture on a production workload.
Multimodal Emotion Recognition
Cross-modal attention model combining HuBERT audio and EfficientNet visual encoders with bidirectional fusion across 8 emotion classes.