Here i have show cased my work from previous internships and projects.
I am currently an Automation Software Engineer at Rivian. I help in redefining what automation solutions in the manufacturing industries can be while automating the process of designing, developing, and deploying new production lines.
I was a Robotics and Controls engineering intern at the Tesla factory in Fremont, California during the summer of 2019. I worked on making a CICD pipeline for FANUC and KUKA industrial robots and also worked on a POC for a teach pendant less robot programming technique.
My work revolved around various domains of machine learning and computer vision such as 6D pose estimation, synthetic data generation as well as Object Detection. I have worked with Nvidia's tools such as Isaac Sim Unity3D Engine, Nvidia Deepstream SDK and Nvidia Isaac SDK. Some of my contributions were making a data annotation tool, as well as implementing domain randomization for generating data. I am also into MLOps and have worked on tools such as Docker and AWS.
I was a Robotics Engineering Intern at Keepsake Automation LLP. I worked on KUKA robots for welding using MIG, TIG techniques. Also helped in setting up cells for industrial robots.
I implemented from scratch a Continuous Integration and Continuous Deployment (CI/CD) pipeline in for FANUC robots. Developed features such as trigger based code backup, remote code push, static analyzer, control flow tool.
Initiated virtual commissioning for auto-trimming Tesla solar panels, using a FANUC robot and Beckhoff PLC in Siemens Process Simulate
Performed thorough market research for the project of the teach pendent-less robot programming. Onboarded a new vendor and orchestrated the project with them for deployment.
Researched on reducing the synthetic to real data gap using domain randomization technique for object detection and pose estimation using Nvidia ISAAC Sim SDK and Unity 3D engine.
Orchestrated a computer vision project from business understanding, custom data annotation, model training all the way to deployment for production using Docker, Kubernetes and AWS
Developed a pipeline to recognise human activity for custom tasks such as loading, unloading. Also deploying the model on NVIDIA Jetson Nano for Edge Inference
Helped in creating an in-house data annotation tool allowing tracking annotated objects using a Siammask tracker. The annotation tool also has a capability of generating labels in KITTI, VOC as well as COCO formats.
Worked with an Azure Kinect camera for developing techniques to track a 3D object using Cumulative Point Feature Matching and Iterative closest point algorithms.
Worked on making a 6D pose estimation model for confirming assembly of parts using vision cameras of Hololens 2