Independent ML engineer in Tiruppur, India, working on text embedding models and evaluation for low-resource languages. Author of morisien-embed, the first embedding model for Mauritian Creole (Kreol Morisien), and of MorisienMTBitextMining, the first Mauritian Creole task in MTEB, the benchmark used across the field to compare embedding models. Also published sante-embed, a 1.5B-parameter model for medical retrieval in nine languages. All models, datasets and training code are released under MIT. Upstream contributor with 18 merged pull requests across sentence-transformers, PEFT, MTEB and Haystack, including fixes to adapter merging and injection in PEFT and to semantic search, ONNX export and evaluation in sentence-transformers. All work to date was trained on a free Kaggle T4.