Sayak Dhar

About Sayak Dhar

I am a Data Scientist with a strong foundation in statistics, machine learning, deep learning and large language models, with experience in engineering, pre-training, and fine-tuning BERT- and GPT-based architectures for genomic sequence modeling and classification tasks. My work spans from literature-driven prototyping and exploratory data analysis to rigorous data cleaning and the design of advanced architectures using techniques such as curriculum learning, contrastive pretraining, and parameter-efficient tuning. I build scalable, distributed cloud pipelines that leverage spot instances for cost efficiency, deploy with MLOps best practices and incorporate real-time monitoring and logging to ensure high performance while optimizing speed and cost. I enjoy collaborating across disciplines, distilling complex concepts into clear insights and delivering presentations that support seamless coordination. With a strong interest in emerging methods - from CNNs and GNNs to foundation and state-space models, as well as multimodal integration, I strive to turn data into solutions that advance scientific understanding and create tangible business value.

Experience

1. Associate Bioinformatics Engineer

   Strand Life Sciences

   Aug 2024 — Present · Bengaluru, IN

   Project 4]: Developing and pretraining a domain-adapted foundation model architecture for predictive modelling.[Project 3]: Fine-tuned LLMs for classification tasks on high-dimensional omics datasets and derived domain-specific insights.[Project 2]: Developed a multi-label deep learning framework for accurate prediction of biological outcomes associated with molecular compounds.[Project 1]: Implemented an end-to-end analysis pipeline for pathogen identification by integrating multi omics data.

Education

• Calcutta University, Kolkata

  Bachelor of Science - BSc, Microbiology Honours

• Indian Institute of Technology, Bombay

  M.Sc., Biotechnology