Experience

AGEYE

• United States
• Farming
• 1 - 100 Employee

• Senior Deep Learning Engineer

  • Aug 2023 - Present

  Bengaluru, Karnataka, India

• Computer Vision Engineer

  • Dec 2022 - Aug 2023

  Bengaluru, Karnataka, India - Developed and implemented computer vision models for object detection, segmentation, and tracking, resulting in a ~25% increase in accuracy compared to previous models. - Demonstrated a 40% reduction in system cost by replacing expensive LiDAR sensors with real-time depth estimation(range: 8-40 cm) and stereovision models with minimal reduction in system accuracy(~5%). - Utilized AWS cloud computing services like EC2 and Lambda for parallel processing and distributed computing… Show more - Developed and implemented computer vision models for object detection, segmentation, and tracking, resulting in a ~25% increase in accuracy compared to previous models. - Demonstrated a 40% reduction in system cost by replacing expensive LiDAR sensors with real-time depth estimation(range: 8-40 cm) and stereovision models with minimal reduction in system accuracy(~5%). - Utilized AWS cloud computing services like EC2 and Lambda for parallel processing and distributed computing, while containerizing pipeline components with Docker to ensure portability and easy deployment across diverse environments.



Tata Elxsi

• India
• IT Services and IT Consulting
• 700 & Above Employee

• Machine Learning Engineer

  • Jan 2022 - Dec 2022

  Bengaluru, Karnataka, India - Created a tool for generating synthetic data using 3d CAD models of an object which reduced both effort and time in capturing real-world data and annotating them. - Played a significant role in developing an Annotation tool using Aruco markers which reduced the data annotating time by 90% and enabled us to generate thousands of annotated images with minimal efforts. - Developed and fine-tuned state-of-the-art algorithms like PoseCNN and EfficientPose for handling texture-less… Show more - Created a tool for generating synthetic data using 3d CAD models of an object which reduced both effort and time in capturing real-world data and annotating them. - Played a significant role in developing an Annotation tool using Aruco markers which reduced the data annotating time by 90% and enabled us to generate thousands of annotated images with minimal efforts. - Developed and fine-tuned state-of-the-art algorithms like PoseCNN and EfficientPose for handling texture-less objects, incorporating color augmentations and regularization techniques (early stopping, drop-out layers) to mitigate overfitting and shape-match losses. Show less