Productionize large-scaled foundational models through SDK and ML portable library with PyTorch and PyTorch Lightning framework for model training, async S3 checkpointing , and embedding generation.
Build product search fullstack website with React, TypeScript, Tailwind CSS along with RESTful API Gateway connecting with various backend services. Automate creation and hosting multiple large-scaled foundation models & LLMs through deploying them to SageMaker endpoints performing real-time inference solutions and KNN retrieval through OpenSearch clusters. Build online LLM chatbots, offline batch inference, and fine-tuning services using AWS Bedrock, Batch, and Step Function.
Manage deep learning compute platform and training infrastructure with AWS Batch and EC2. Build backend ML data pipelines, CI/CD, release management, and distributed deep learning framework with end-to-end support for multiple engineering and science teams across Amazon.
