Soma Meghana Prathipati

About

ML/robotics engineer with Apple experience who built a computer-vision-driven industrial defect detection system integrating a robotic arm with ROS-based real-time inference on an edge GPU. Drove major performance gains (cut inference time ~60% via quantization + TensorRT) and improved robustness to lighting/material variation, with strong emphasis on production reliability (health checks, watchdogs, observability, CI/CD) and interest in shaping early-stage startup engineering culture.

Key Strengths

• End-to-end build of real-time vision + robotics defect detection system (data collection through deployment)
• Inference latency optimization via quantization + TensorRT (reduced inference time ~60%)
• ROS-based real-time integration between vision, inference, and robotic control (publisher/subscriber, actionlib)
• Production reliability engineering: profiling end-to-end latency, asynchronous callbacks, watchdog timers, health checks
• Robustness to domain variability (lighting/material texture) using adaptive preprocessing and augmentation
• Edge deployment and containerization for low-latency pipelines (Docker on edge GPU)
• Scalability mindset informed by Apple experience (early profiling, iterative optimization, resilience checks)

Experience

Education

Skills

Languages