Omeed Jamali

About

Robotics/AV-focused engineer who built an end-to-end gesture controller for a GEM e2 autonomous vehicle using YOLOv8 pose and ROS, including model training, ROS perception nodes, and a safety-oriented state machine (stop override + hold-to-register). Also has internship experience at Intramotev integrating LiDAR object detection via Redis pub/sub and performing sensor-frame calibration (roll/pitch correction using ground-plane normals), plus Dockerized deployments and Gazebo-based testing.

Key Strengths

• Built end-to-end gesture-control pipeline for an autonomous GEM e2 vehicle using YOLO pose + ROS
• Improved gesture classifier robustness by adding an idle class and tuning confidence thresholds
• Designed integration logic/state machine where stop overrides actions and other gestures require 3-second hold
• Implemented ROS perception node subscribing to RGB/depth topics and publishing model outputs for control
• Reduced message load by publishing only state changes instead of continuous classifications
• Debugged real-time LiDAR alignment by correcting roll/pitch using ground-plane normal calibration
• Used Gazebo to validate vehicle motion/controller behavior when physical testing wasn’t available
• Containerized ML/robotics environment with Docker to make setup reproducible for other users

Experience

Education

Skills

Languages