Somya Das

Machine Learning Engineer with 2.5+ years of experience building production computer vision and ML solutions for life sciences. Specialized in Generative AI, LLMs, and Retrieval-Augmented Generation (RAG), with hands-on experience developing end-to-end AI applications using FastAPI, LangChain, Qdrant, Pinecone, PyTorch, and local LLMs. Skilled across the full ML lifecycle, from data pipelines and model development to Dockerized deployment and REST APIs, with a strong computational biology background.

Software Engineer I — Machine Learning & Computer Vision - Thermo Fisher Scientific - Bengaluru, India

(2022-01 - 2024-07)

• Developed production ML and computer-vision pipelines for scientific-imaging software used in life-science research workflows.
• Built ML pipelines to predict subcellular protein localization from multi-channel fluorescence microscopy images, combining image segmentation, feature extraction, UMAP dimensionality reduction, a pre-trained Variational Autoencoder (VAE), Kernel PCA, and k-Nearest Neighbors (k-NN).
• Developed automated cell-counting algorithms using connected-component analysis, improving counting accuracy by ~40% over the manual baseline.
• Created a cell-line recommendation system from biological metadata and experimental inputs, with Python data pipelines integrated with AWS S3.
• Partnered with R&D, Software, QA, and Product teams in Agile sprints; authored technical documentation and presented project outcomes to scientific and business stakeholders; delivered milestones on schedule and was recognised by manager for problem-solving and project contribution.

Post Graduate Diploma - Big Data Biology - Institute of Bioinformatics & Applied Biotechnology (IBAB), in collaboration with IIIT Bangalore (2021-01 - 2022-01)

M.Tech - Biological Sciences & Bioengineering - Indian Institute of Technology Kanpur (2018-01 - 2020-01)

B.Tech - Biotechnology - Dr. A.P.J.

Abdul Kalam Technical

University (2014-01 - 2018-01)