PromoterFinder
A privacy-forward dApp for machine learning on DNA sequence
Transformers are increasingly capable tools for genomics, but deploying them raises real privacy concerns around sensitive DNA data. In this project, I fine-tuned a mini-DNABERT model for proximal and core promoter region identification, then built a dApp on the Oasis Blockchain to host it — enabling users to run predictions on their own sequence data in a confidentiality-preserving manner. The project also benchmarks Cipher ParaTime vs Parcel-hosted models for private ML inference.