Shubhankar Joshi is an AI Associate Scientist at Genentech in the San Francisco Bay Area, California. With 2 years of experience, his work focuses on technical architecture, particularly in architecting high-throughput bioinformatics pipelines using Nextflow, SLURM, and Airflow for large sequence datasets. He has developed Transformer models in PyTorch for RNA–protein interaction prediction and deployed a RAG microservice on Google Cloud Run. Shubhankar also utilizes PyTorch Lightning, FlashAttention, and Weights & Biases for distributed training, and builds full-stack tools with TypeScript and React. He is interested in agents and MOE and enjoys knowledge sharing with peers.

Agents, MOE

At Genentech, current work involves architecting high-throughput bioinformatics pipelines using Nextflow, SLURM, and Airflow for large sequence datasets. Key projects include developing Transformer models in PyTorch for RNA–protein interaction prediction and deploying a RAG microservice on Google Cloud Run. Technical activities involve utilizing PyTorch Lightning, FlashAttention, and Weights & Biases for distributed training, alongside building full-stack tools with TypeScript and React.