Turning text to Alpha for the Investment Industry.

1. MLOps using Kubeflow – Developed a scalable object detection ML workflow using the Faster R-CNN model for launching training, serving, and scaling models on Kubernetes.

2. Developed a Pipeline in Python using Kubeflow’s Pipeline component to demonstrate the usage of object storage (MinIO) to store training data and models instead of the file system – Aiding in scalability and faster data retrieval.

3. Developed a solution for deploying Kubeflow on an on-premise Kubernetes cluster using Helm charts.

4. Analytics Platform as a Service – developed a one-stop solution using Docker, Kubernetes, and Helm to deploy major analytics components (NiFi, Kafka, Kubeflow, Spark, MariaDB and Cassandra) on Openshift Kubernetes for all the Data Analytics needs.

1. Designed and developed a major feature in Java for NiFi to expose its metrics to Prometheus. Made this feature Open Source: https://github.com/apache/nifi/pull/3257.

2. Performed a POC on Argo Workflows and demonstrated its capabilities to the team. As part of the POC created a workflow to read and write data from the Hadoop file system.

3. Implemented an automation test suite for NiFi by generating ~3 GB of CSV data in the file system and sending ~45 MB/s of CSV data to NiFi using Python, Ansible and Jenkins tools.

4. Designed and Developed a feature in Java for Apache Beam to read and write data from Cassandra. Design Doc: https://github.com/apache/beam/files/6283847/Cassandra.4.x.upgrade.docx – Aiding in increased security.