Avinash Chikkadlur Ravi Shankar

Worked on building software infrastructure for Automated Valet Parking (AVP-2) systems 1. Optimized a multithreaded software infrastructure for image acquisition and processing at remote facilities, which resolved memory leakages and access violations due to data race conditions 2. Developed video encoder and decoder tools to record remote camera streams over RTSP for internal data analysis and neural network training Skills: Software Infrastructure, Gstreamer, Rust, C++, Software Development

Worked on developing the next generation of smart AI-based dishwashers for Miele. My roles included developing software component architectures and integrating deep-learning-based object recognition features that assist and improve user experience with an intelligent selection of programs. Skills: Software Integration, Software Architecture, Python, Powershell, Embedded Software Programming, C++, Continuous Integration, Unit Testing

Worked as an external software engineer for ADAS at Robert Bosch. 1. Developed a tensor library in C++ for inference on ARM SoCs, conforming to MISRA C++ standards 2. Prepared and augmented image datasets, trained and deployed deep-learning models for free-space detection on automotive ECUs 3. Optimized inference algorithms to compute fixed-point data to reduce model size 4. Optimized algorithms with SIMD instructions for faster computation on embedded SoCs Skills: C/C++, Assembly, Python, Deep Learning

Topic: Convolutional Neural Network-based Static Landmark Recognition in Urban Areas 1. Surveyed suitable deep learning models for static landmark recognition 2. Designed and trained models for semantic segmentation of objects in urban areas 3. Investigated models trained with RGB and RGB-D (with depth from stereo cameras) images for higher classification accuracy

Topic: Algorithm development to generate point-cloud for static objects with stereo cameras 1. Developed interfaces to integrate POE stereo cameras with OpenCV library for calibration 2. Implemented stereo block matching algorithm in OpenCL for faster depth-map rendering

Developed ADTF plugins for SIFT and SURF algorithms in OpenCV for feature extraction from images.

Automotive Software Engineering, Machine Learning, Computer Vision

Electronics and Communication Engineering