Swapniel worked on core features for the deepset-ai/haystack and haystack-core-integrations repositories, focusing on backend and full stack development using Python, Haystack, and Weaviate. He delivered a hybrid retrieval system by introducing the WeaviateHybridRetriever class, enabling combined keyword and vector search within the Weaviate document store to improve document relevance and search efficiency. Additionally, he enhanced the embedding pipeline by adding a revision parameter to all Sentence Transformers embedder components, allowing users to pin model versions from the Hugging Face Hub. These contributions improved reproducibility, stability, and user control across search and machine learning workflows.