Saumya Das

I’m a Machine Learning Engineer with ~2.6 years of experience in computer vision, image processing, and applied ML for healthcare and life sciences. I build end-to-end pipelines—from data processing and model integration to analysis—using Python and deep learning frameworks.

At Thermo Fisher Scientific, I delivered an end-to-end Python pipeline for fluorescence microscopy protein localization prediction, including OpenCV-based preprocessing, DAPI-seeded watershed segmentation, and VAE feature extraction with kernel-PCA and k-NN. I deployed the inference pipeline on AWS (EC2 + S3) and fixed production concurrency and disk-pressure issues with multithreading. I also developed ML-based cell line recommendation prototypes, collaborating with R&D, product, and QA teams while keeping documentation updated in Confluence and tasks tracked in Jira.

Today, I’m expanding my expertise in Generative AI, LLMs, and RAG systems, and I enjoy translating complex ML concepts into practical workflows for others. As an AI/ML Trainer (self-employed), I design hands-on workshops for biotech students and professionals, and I’m building GenAI healthcare PDF Q&A using LangChain and Pinecone.