Similarity search is a well-studied task aimed at identifying data with common elements. As data size grows, this process becomes increasingly complex and computationally demanding. In this work, we focus on finding sequences similar to a user-provided sequence query. Most approaches rely on applying a similarity function to each sequence, which becomes inefficient with large datasets. To overcome this, we introduce SISIS, a method that leverages sequence indexing for faster retrieval of sequences with matching elements in the same order. Furthermore, to provide similar sequences even when their points do not exactly match but are contextually close, we introduce SISIS*, a variant of SISIS based on points embeddings. This variant yields more comprehensive results than the exact SISIS method. Our extensive experimental validations demonstrate that the proposed approach significantly outperforms a baseline method based on the well-known LCSS (Longest Common SubSequence).