Iditarod BioData Scientist (Software Developer)
Sep. 2024 - Jun. 2025Brooklyn, New York, United States- Optimised the Python code and reduced the pipeline execution time by 57%, which also resulted in AWS cost reduction of 53%.
- Implemented compression strategy for large files and configured CloudFront that reduced the loading time of the website from 30 seconds to 800ms.
- Built an ensemble of binary classifiers trained on a million protein structures to distinguish wild-type proteins, then used transfer learning on their embeddings to predict solubility-enhancing mutations.
- Built a vision algorithm on 3D dataset along with integrating existing models into scalable, production-ready systems using AWS.
- Worked with CNNs, MLMs and Diffusion models.