Kapil Punamchand

OVERVIEW
Led architecture and development of an automotive 360° top-view system on TI's TDA2ex SoC with four vision cameras, delivering surround views and ADAS-integrated guidelines and artificial intelligence application porting and optimization. I directed architecture and design, establishing a robust ADAS framework.

KEY RESPONSIBILITIES
• Developed Ethernet communication modules for sensor interfacing, targeting ARM Cortex A15 CPU for back-channel communication.
• Designed AUTOSAR-compliant software components and complex device drivers (CDD).
• Created video pipelines using RTP, MPEG-TS, and H264 codecs.
• Gained experience with the H264 codec and configuration to meet required quality and bandwidth requirements.
• Integrated and benchmarked a mobilenet1.0 and mobilenet 0.5 ONNX model on the TI TDA2ex DSP, developed a video pipeline with pre-processing, deep learning inference, and post-processing output sent over CAN.
• Converted the ONNX model to a TFLite model to be portable on ARM cores and benchmarked it.
• Implemented and validated time synchronization using PTP and NTP on a Linux-based ECU system.
• Worked on TI TDA4 OpenVX and GStreamer-based video pipeline.

OVERVIEW
Developed video encoding/decoding over IP on ICS-8580 using a transport stream (TS) for video, audio, and metadata. I enhanced Boeing 747-8 software with integrated aircraft data for ACARS downlink reporting.

KEY RESPONSIBILITIES AND ACHIEVEMENT MODULE DEVELOPMENT
• Analyzed requirements and developed C modules in Linux to enable/disable video/audio transmission over IP.
• Enhanced and maintained MPEG-TS muxer compliant with ISO 13818 for multiplexing video, audio, and metadata.
• Enhanced 747-8 flight software with ACARS downlink headers for data like position and speed, supporting aircraft tracking and safety.

Spearheaded a team of six, handling task delegation, technical support, and project delivery. I managed client communication, weekly progress updates, and aligned project goals with client expectations.

KEY RESPONSIBILITIES
• Developed TI Vision SDK components, optimized DSP modules, and ported software from Renesas to TI TDA3x platforms.
• Gained experience with Lidar, Radar, GPS, and IMU sensors for autonomous driving and participated in sensor fusion, path planning, and motion planning requirements.
• Designed video pipeline use cases to cater to the needs of different ADAS applications using the vision SDK use case generator.
• Attained target performance through DSP intrinsics, DMA/EDMA, and L2 memory optimization.
• Designed Pan and Zoom module for camera monitoring systems (CMS).
• Developed a multi-threaded video framework with Linux semaphores on the NVIDIA TX1 platform and integrated different face and eye monitoring algorithms to decide driver readiness.