Experience💼

At Rivian, I led the development of a computer-vision–based motion perception system for tuggers and AGVs in collaboration with Petra AI, Rivian’s Manufacturing AI team. Using dense optical flow (MemFlow / RAFT), the system classified motion states, detected stop–start events at millisecond resolution, and distinguished vehicle ego-motion from surrounding activity. I designed a hybrid edge–cloud pipeline, combining custom dual-camera capture with CUDA-accelerated PyTorch inference on AWS, to automatically generate downtime classifications and process-efficiency insights from raw video data.

In parallel, I developed a real-time vision-guided robotic manipulation pipeline using Keyence multi-camera systems and FANUC robotic arms for production-scale part handling. This work focused on translating camera observations into reliable robot motion through 3D object localization, camera-to-robot spatial transformations, and trajectory generation, enabling precise and repeatable pick-and-place operations under dynamic shop-floor conditions.

Alongside these perception-focused projects, I served as Controls Integration Lead for the R2 Front-End Module (FEM) assembly line, taking the system from design through production readiness. I worked across controls architecture, PLC/HMI validation, safety systems, and MES/MARS integration, while building the digital quality thread that ensured full genealogy and traceability across scans, vision checks, torques, and continuity tests on a high-throughput production line.

Developed a real-time computer vision system using Python and OpenCV for object detection, screw alignment, and robotic path guidance in an automated screwing application.

Integrated the perception pipeline with Beckhoff PLCs via TwinCAT v3.1 and HMI interfaces, enabling seamless coordination between the vision system, KUKA industrial robots, and servo-driven actuators.

Enabled precision control by interfacing with DH Robotics grippers, inclinometers, and temperature sensors using industrial communication protocols such as EtherCAT and Modbus.

Contributed to improving overall system accuracy and efficiency in a high-speed advanced manufacturing setup by bridging robotics, perception, and control layers.

Developed design deliverables for instrumentation and control systems in oil and natural gas projects.

Created and verified piping and instrumentation diagrams (P&IDs).

Managed the commissioning of instrumentation systems, ensuring proper installation and configuration.

Verified field instrumentation locations and conditions.

Worked with software tools such as SP3D, SmartPlant Instrumentation, and Microsoft Excel.

Developed cutting-edge algorithms utilising 2D LIDAR technology for electronic toll collection, speed monitoring, enforcement, and vehicle profiling.

Achieved significant efficiency gains with a 22% reduction in toll processing time and a 40% cost reduction by strategically choosing an array of 2D LIDAR sensors over 3D counterparts. Employed a 2D bitmap generation for comprehensive vehicle record keeping and maintenance.

Implemented state-of-the-art LIDAR SLAM algorithms, notably Hector SLAM, enabling precise real-time mapping and vehicle tracking using Robot Operating System (ROS 2).

Addressed critical aspects of scalability, maintenance, and regulatory adherence, making the system adaptable for nationwide deployment.

Integrated Heterogeneous control system for a manufacturing facility to AWS cloud:

Learned Fundamentals of Labview Basics, Digital Twin, IIoT, noxVIEW DFI. Observed the Process and Factory Automation.