Highlights 🔗

• Adapted and enhanced Deepsort algorithm for tracking dynamic objects

Highlights 🔗

• Estimated Lidar to Lidar extrinsic calibration matrix using autoware software

Highlights 🔗

• Estimated Lidar to RGB camera extrinsic calibration matrix using autoware software

Highlights 🔗

• Estimated Thermal camera intrinsic calibration parameters using open-cv python library

Work Experience 🔗

Full-time

Cincinnati, Ohio, US

Oct 2021 to Mar 2022

Highlights 🔗

• Developed FLIR thermal camera grabber ROS Node for perception stack
• Developed Lidar segmentation, Lidar Tracker and Sensor fusion algorithm for autonomous tractor
• Worked on Lidar ground segmentation algorithm by following state-of-art research papers
• Worked on enhancement of PointPillar and mmdetection3d Deep learning models for 3D Object detection
• Trained and deployed Yolo v4 model for detection of pedestrian, Cones, K-Loader and airplanes
• Worked on sensor fusion of 5 RBG cameras and 2 Lidars

Highlights 🔗

• Developed Flir Thermal camera grabber ros node for Perception stack

Highlights 🔗

• Developed Ansible scripts to validate Linux VM capacity and installation prerequisites
• Provisioned Linux VM in Cloud environment using automated scripts
• Performed post-provisioning activities such as IAM, LDAP integration, updating security roles, filesystem creation and Linux patching
• Used Ansible to install prerequistes and infra components in all 12 VMs including JAVA JRE, Postgresql, Hadoop ecosystems and Neo4j database
• Used Ansible to install java application, creation of services, update configurations and installation of Elasticsearch, Logstash and Kibana
• Performed deployment validation checks using Ansible powered tool called Enigma

Highlights 🔗

• Reduced duplicate alerts by 80%
• Smart handling of alerts by using Probable cause analysis to find the root cause and fix them, ignoring other dependant tasks
• Created adapters to integrate ignio with ServiceNow, Solarwinds and AppDynamics using REST API
• Created dynamic profiling of systems, VMs, Applications, Databases, Network and storage entities
• Using dynamic profiles, predicting if the alert is valid for the given conditions
• Auto verify and apply self-healing scripts to address those alerts
• Automated more than 90% of repeated IT Alert and hence enabling technical team to work on solutioning rather than monitoring and troubleshooting