Saravanakumar A

About Saravanakumar A

Currently, I am a technical lead at Tata consultancy services, India. I have 14+ years of experience in automotive (ADAS,Autonomous cars) and broadcast industry. I have worked in the field of deep learning, computer vision, radar signal processing, audio compression and video compression. Furthermore, I have also ported and optimized the above algorithms in various DSP platforms like Texas Instruments, Analog devices, and Qualcomm. I have contributed many projects for various geographic clients with critical delivery condition as well. At present, I am more passionate to research and develop artificial intelligent-based solutions. Development of Deep Learning Algorithms(CNN based networks - model creation, fine tunning, Data ingestion, Data preprocessing, Data augmentation etc.) Development of Computer Vison Algorithms(Feature Implementation using Machine Learning-EHOG,2D Construction Using Stereo Camera BM-SGBM,Tracking of Objects-Sparse Optical Flow-Kalman Filter,Surround View camera Image Stitching,Image Enhancement-2D filter design and Clustering of Images.etc.) Development of Radar Signal Processing Algorithms for 77GHz Multimode Radar( RSP Feature Algorithm Development : Lane change Assit-LCA,Front Cross Traffic Alert-FCTA. Knowledge in RSP Layers : Range FFT Calculation, Doppler FFT, Digital Beam forming. Tracker layers : SR MR fused tracker, Kalman filter) Porting of Computer Vision Algorithms(TDA2x : DSP,EVE,A15 Frame work: Link Chain),Interfacing Hardware(Stereo Camera,Omni Vision Mono Camera with TDA2x) Hands on Experience on ISO26262 functional Safety Software Development of Compression Algorithms( Audio : Multi Channel HD Audio compression with Post Processing -TrueHD,AC3,DPL2Z. Video : H.264 baseline Multi resolution HD Encoder - Verified buffer overflow ness, constant and variable bit rates of algorithms. Speech : EVRC-B) Porting of Compression Algorithms( Audio : Analog Devices processors -BF533,BF561 & Sharc with AVR Frame Work. Video : TI Davinci DM6467 with Codec Engine Frame work. Speech : QDSP6) Tools used: Programming languages: C, basic C++, Python, Hardware Assembly languages (Ti, Analog Devices). Deep Learning Frameworks,IDE & Libraries: Tensorflow,Keras,OpenCV,matplotlib,Anaconda,Jupyter. Processors: Ti DSP(DM6467, DM814x,TDA2x), Analog Devices(Blackfin, Sharc),Qualcomm(QDSP6),NVIDIA. Other tools: Hardware RTOS(BIOS, VDK),Hardware IDE(CCS, hexagon, VDSP),MSVC,OS(Linux, Windows).

Experience

1. Technical Architect(Adas-advanced Driver Assistance)

   Tata Consultancy Services

   Feb 2013 — Present · Bengaluru, IN

   Developing Deep Learning, Computer Vision, and Radar signal processing feature Algorithms(Driver facial emotion detection, Front Cross-traffic alert using RSP, Lane change assist using RSP, Traffic Light Detection, Traffic Sign recognition, Pedestrian Detection, Vehicle Detection, Surround View Monitoring,Object Tracking,Stereo Calibration,Disparity algorithms and Distance Measurement),PreProcessing(Image Enhancement,Clustering). Porting these algorithm in TDA2x (optimise it on C66x and Embedded Vector Engine(EVE)) and Davinci DM6467 of Texas Instrument Processors(TI) and integrating Stereo Camera in TDA2x Functonal Safety ISO26262 software for Autoliv airbag systems

Education

• Sona College of Technology

  Bachelor of Engineering (BE), Electrical and Electronics Engineering

• College of Engineering, Guindy

  Master of Engineering(ME), Applied Electronics