profile-pic

Guruprasad Jadhav

An accomplished Senior Software Engineer with a strong passion for Data Analytics and Machine Learning. I take pride in utilizing my technical skills and expertise to create innovative solutions that drive real-world impact.

My journey in the field of technology began with a Bachelor's degree in Computer Science and Technology from Shivaji University, followed by a Master's in Data Analytics from the prestigious National Institute of Technology. During my academic journey, I developed a solid foundation in programming languages like Python, C, C++, and SQL, as well as cutting-edge frameworks such as Scikit, TensorFlow, Keras, and NLTK.

As a professional, I have been fortunate to be part of exciting projects at Visteon Corporation, where I have honed my abilities in AI on the Edge and resource-constrained device deployment. My experiences in developing an IR camera-based driver assistance system and integrating Android applications with Qualcomm SA8155 ADP Board for Surround View Monitoring have strengthened my technical skills and problem-solving capabilities.

  • Role

    Sr. Software & OpenGL Engineer

  • Years of Experience

    4.8 years

Skillsets

  • NLTK
  • Vision transformer
  • NLP
  • ADB
  • Windows
  • web
  • TensorFlow
  • SQL
  • Scikit
  • Python
  • OpenGL
  • OpenCV
  • Android
  • MySQL
  • Machine Learning
  • Linux
  • Keras
  • Git
  • Dlib
  • Deep Learning
  • Computer Vision
  • C++
  • C

Professional Summary

4.8Years
  • Jan, 2021 - Present4 yr 10 months

    Senior Software Engineer

    Visteon Corporation

Work History

4.8Years

Senior Software Engineer

Visteon Corporation
Jan, 2021 - Present4 yr 10 months
    Part of the Driver monitoring system project team. The objective of this project team is to develop an IR camera-based driver assistance solution using deep learning, that runs on low-cost processors. Data collection of various real scenarios with IR camera and optitrack camera. Preprocess the data of cleaning, Transformation, and Reduction by using various techniques. Proficient in Python and C++, solid foundation in algorithms and data structures. Worked on AI on the edge, experience in running AI models on resource constrained devices. Using Optimization techniques like Pruning and Quantization to reduce the complexity of the architecture and increase the performance of the model. Train neural networks in Python and perform inference in C++. Having knowledge of end to end DMS architecture. Designed and implemented a cross-platform Driver Monitoring System (DMS) on Qualcomm SA8775 and SA8295 platforms, targeting both Linux and QNX environments. Collect feedback from QA team and product managers, analyze and prioritize feedback that will be worked upon. Mentor team members and help them solve hard problems.

Major Projects

6Projects

Face Detection

    Developed and optimized a real-time face detection system using an Embedded SSD MobileNet v1 architecture tailored for edge devices. Designed to accept both full-resolution and ROI images with no preprocessing steps. Technologies used: Python, C++, OpenCV.

Brightness Control System

    Automatic set gain and exposure value in the IR camera to control brightness, increasing the accuracy of other DMS deep learning models. Trained a multivariate polynomial regression to determine gain and exposure value with respect to brightness. Technologies used: Python, C++, OpenCV.

Face Recognition

    Implemented a lightweight, high-accuracy face recognition system based on the hybrid CNN and Vision Transformer EdgeFace architecture, optimized for real-time inference on edge devices. Designed to register and recognize individuals from a single input frame. Technologies used: Python, Dlib, OpenCV.

Liveness Face Detection

    Developed a model for liveness detection to spot fake faces and perform anti-face spoofing in face recognition systems. Trained to distinguish between genuine human faces and fakes, enhancing security efficacy. Technologies used: Python, Scikit-learn.

Fine-tuning the DMS models for different camera positions

    Fine-tuned face and eye detection models for the A-pillar camera position. Retrained SSD mobile architecture on collected IR images and used DLIB and OpenCV for ground truth generation. Technologies used: Python, Tensorflow.

Head Pose Estimation

    Predicted drivers' head pose angles (Pitch, Yaw, Roll) to check distraction. Trained a CNN for regression to determine 3D head pose angles and applied model optimization techniques to reduce size and improve inference speed. Technologies used: Keras, Python.

Education

  • Master of Technology - Data Analytics

    National Institute of Technology Trichy (2020)

  • Bachelor of Technology - Computer Science and Technology

    Shivaji University Kolhapur (2018)