BIOGRAPHY

Hi, I’m Mukul Ganwal - a passionate roboticist driven by my love for robots and their potential to transform our world.

Here’s my Curriculum Vitae for detailed information.

I am a Perception and MLOps Engineer at Anyware Robotics, where I lead a 4-person team developing end-to-end robotic palletizing systems that integrate computer vision with industrial automation. My work focuses on building production-grade machine learning pipelines that enable robots to understand and interact with complex real-world environments. Currently, I’m shipping systems that use RGB-D cameras and advanced perception algorithms to reconstruct work cells in 0.1 seconds, allowing FANUC robots to palletize boxes with 6.3-second cycle times.

My expertise lies at the intersection of robotics, computer vision, and MLOps. I specialize in creating active-learning workflows that dramatically reduce data labeling requirements - achieving 40% reductions in manual annotation while maintaining rapid 40-minute test-to-deploy cycles. I’ve pioneered the use of synthetic data generation through Blender photogrammetry combined with StyleGAN3 domain randomization, scaling datasets by 8× and improving detection performance by 4.7 percentage points. My systems integrate seamlessly with ROS 2 architectures, incorporating behavior trees and hierarchical finite-state machines with real-time fault classification that has reduced unplanned robot downtime by 35%.

Before joining Anyware Robotics, I honed my skills as a Graduate Research Assistant at the BioRobotics Lab in Pennsylvania, where I engineered modular battery extraction mechanisms for Apple devices and built real-time multi-class instance segmentation pipelines achieving 98% accuracy. My journey into robotics and AI began with internships at PGP Glass and other companies, where I developed visual inspection systems with 99% defect detection accuracy and created intelligence systems that transformed business operations.

My technical toolkit spans PyTorch, TensorFlow, Ray, Keras, JAX, CUDA, and ROS/ROS2 for ML and robotics, complemented by Docker, Kubernetes, GitHub Actions, Terraform, AWS, and GCP for robust MLOps deployments. I’m proficient in Python, C/C++, MATLAB, and Bash across Ubuntu, Debian, and FreeRTOS environments, with expertise in simulation platforms like Webots, Gazebo, and V-Rep alongside CAD tools like SolidWorks.

What drives me is the belief that robots can fundamentally improve how we work and live. Every system I build bridges the gap between cutting-edge AI research and practical robotic applications that deliver measurable impact in real manufacturing environments.

“My heart in my work” - Andrew Carnegie