Vivek Tangudu

Designed and implemented high-performance C++ components for a real-time 4K image-streaming system using a modular, multi-threaded architecture. Built a custom zero-copy streaming protocol that shares a DMA buffer between camera and networking layers to minimize intermediate copies and reduce end-to-end latency. Reduced latency by 25% and decreased memory consumption by replacing per-frame allocations with persistent zero-copy frame buffers. Diagnosed bottlenecks using packet inspection, GDB debugging, and CPU/memory profiling across the capture-to-network pipeline. Implemented ONVIF-compliant communication using SOAP/XML for discovery and configuration, and built a backend-switching abstraction layer with controlled shutdown and re-initialization to keep the control plane stable.

Developed a cross-platform gimbal–camera emulator and SDK in C++ to support integration and testing on Linux and Windows. Re-engineered the client-to-gimbal command path to sub-10 ms latency using a priority-queue scheduler that preempts latency-critical commands. Performed protocol debugging and integration testing to validate communication between the emulator, SDK, and hardware devices across both operating systems.