Hire Rajesh S. - MLOps Engineer

1. Building and maintaining the Ray cluster with custom docker image and deploying the solution using Kuberay operator in k8s by using Flux CD system as part of ML platform team. Handling the scalability of the ray application based in the kubernetes pods. 2. Taking care of JupyterHub & Apache Airflow 3.x in a multi cluster environments & multi workers setup. 3. Actively designing and working on to setup mlflow with multi tenant capability.

1. Built Automated deployment training pipeline for ML models using mlflow, boto3 scripts to create AWS ECS service enabling the online inferencing pipeline. Goal is to develop the platform for DS and Business owners for quickly deploy the model on their own. 2. Designed the One click deployment through UI enabling self-serving platform for Data scientist to deploy model on their own. Implementation is in progress. 3. Designed AB testing strategies for Fintech models using Launch Darkly feature. 4. Designed and implemented Automated Elevation for the ML model through schedule basis using Lambda function and Schedule Rule invocation. Working on implementing other Model elevation strategies in coming days. 5. Deployed models which scales as per the number of input request. Ex: The API can handle 70 RPS for fraud detection model at peak hours which is 0.25 million request per hour. Scaling policy is applied as per the CPU utilization of the ec2 instance.