Hi there!

I'm Sai Mukkundan Ramamoorthy (Sai M. Ramm),

M.Sc. Autonomous Systems | Software Engineer | 3D Computer Vision & Robotics

I am a Software Engineer with 6+ years of production-grade experience in C++, embedded software, and firmware development, currently transitioning into advanced robotics and AI.

Currently pursuing my M.Sc. in Autonomous Systems in Germany, my focus lies at the intersection of low-level hardware performance and high-level AI perception. I specialize in building highly optimized, scalable software pipelines for 3D computer vision, learning based robotic manipulation, and autonomous navigation.

What I'm currently working on:

• Master's Studies: M.Sc. Autonomous Systems at Hochschule Bonn-Rhein-Sieg (HBRS), Germany.
• Teaching: Teaching Assistant for Software Engineering for Robotics, mentoring Master's students in C++, ROS2, and CI/CD best practices.
• Research & Development: Building transformer-based multi-view panoptic segmentation models and visuomotor policies for robotic manipulation.

Skills

3D Vision & Deep Learning:

• PyTorch Transformers OpenCV Open3D / PCL COLMAP Structure from Motion (SfM) skrl (reinforcement learning)

Robotics & Simulation:

• ROS2 IsaacLab / Isaac Sim PyBullet Gazebo Monte Carlo Localization (MCL)

Systems & Embedded Programming:

• Modern C++ Python C Shell Scripting Java
• Linux Kernel Android OS HAL/HIDL RTOS CMake Docker GDB

🚀 Highlighted Projects

(Check out my pinned repositories below for code and visual results!)

• Panoptic-MapAnything: A transformer-based feed-forward model integrating MapAnything with Mask2Former for multi-view consistent panoptic segmentation on the ScanNet++ dataset.
• Imitation Learning-Based Visuomotor Policy: Visuomotor policy development for a Franka Emika arm using behavior cloning and RL in IsaacLab.
• 3D Occupancy Grid Mapping: A high-performance spatial mapping algorithm written entirely in C++ processing LiDAR point clouds.
• Reinforcement Learning-Based Path Planning: Custom implementations of PPO and Q-learning for dynamic obstacle avoidance in a GridWorld.

---

💡 Fun Fact: I spent years optimizing boot times and OTA updates for millions of Android devices before diving into the world of autonomous robots!