Machine Learning Engineer
Current- Designed and developed dynamic deployment APIs, enabling on-demand deployment creation and automated deactivation based on user activity.- Created a Python module as an MLflow plugin to implement Single Sign-On (SSO) foruser authentication and authorization, enhancing security and user management.- Developed Seldon-Core based model deployment framework, ensuring adherence to Python standards.- Dockerized computer vision models and deployed in Kubernetes for scalable and efficient execution as REST service.- Implemented MLflow server using ECS, ECR, and Fargate, enhancing deployment ecosystem efficiency.- Reduced Docker image sizes by up to 80% using the MLflow server, improving scalability and resource utilization.- Designed Argo workflows to orchestrate inference tasks and fetch results efficiently.- Created APIs for real-time model information retrieval, seamlessly integrating them with an in-house website.- Built Grafana dashboards incorporating custom metrics for unified performance insights.- Innovated a tool reducing image data transfer time to S3 by 70%, using multithreading and S3 multipart upload.
