Srinivasa Satya

Deployed 12 Lavo surface-cleaning mobile robots for beta customers and 2 Lavo Refresh robots for production using ROS2, Isaac ROS, and Docker. Owned perception, navigation, and localization stack with cuVSLAM-based VIO and EKF fine-tuning, and developed PX4 firmware in C++ for the Sherpa industrial cleaning UAV.

Deployed YOLOv8/YOLOv11 real-time threat detection with DeepSORT tracking at 86% accuracy, optimized for Jetson Orin via TensorRT. Engineered a GPS-denied drone landing system on ARK Jetson using VIO and adaptive camera pipelines.

Built low-level control for UAV-UGV multi-agent teams with TSP-optimized planning and dynamic obstacle avoidance. Established ground-truth positioning using motion capture systems and developed a ROS2 SLAM pipeline with RTAB-Map and Intel D455.

Built a hybrid LSTM-CNN classifier for rainfall intensity at 92% accuracy from 3D LiDAR and disdrometer data. Generated high-fidelity rain-particle LiDAR simulations in Blender and solved SUMO to Unreal Engine 4 georeferencing.

Engineered real-time WiFi outage detection for in-flight airline connectivity networks using Splunk and Python, cutting anomaly detection latency by 40%. Streamlined AWS log-analytics workflows for cross-airline troubleshooting.

Built YOLOv5 real-time underwater fish and feed detector at 60.5% mAP, optimized via ONNX and TensorRT to 18 ms inference on Jetson edge. Improved underwater visibility with CLAHE and color-correction preprocessing.

Built autonomous UAVs integrating u-blox GPS, Ardupilot, and Pixhawk flight controllers. Implemented a 3D LiDAR-LOAM pipeline producing dense point clouds for VR visualization.

Grade: 3.7

M.S. in Robotics from Northeastern University, completed in December 2024.

B.Tech in Electronics and Communication Engineering from SNIST, completed in June 2020.