Akash James

About

Exploring and pushing the bounds of Computer Vision; Making it available to the masses by democratizing it. I'm very enthusiastic about bleeding-edge technology and love to tinker with the technology of tomorrow. My goal is to make technology usable, something that can make a difference to this planet of ours.

AI Architect & Computer Vision Specialist

  • Birthday: 20th July, 1996
  • Phone: +91 8971358197
  • City: Bangalore, India
  • Age: 24
  • Degree: Bachelors of Engineering (CSE)
  • Email: akash.james2096@gmail.com

I write blogs when I don't code and beatbox pretty much all the time. Working on new things gives me a rush and being in a state of rush is something I am addicted to!

Skills

My expertise lies predominantly in Artificial Intelligence and Computer Vision. My skill-set includes GPU acceleration, Neural networks design, High-Performance Inference Pipeline architecture and Embedded inference devices. I have demonstrated knowledge in DeepStream Framework, TensorRT, PyTorch, Onnx, OpenCV and other Deep learning frameworks utilized in Computer Vision.

Pytorch/TensorFlow 90%
OpenCV 85%
Deepstream 95%
TensorRT 75%
Flask/Celery 75%
C++ 85%
Python 90%
Rapid Prototyping 95%
Team Management 80%

What I have been Upto

A sneak peek into how life has been progressing for me. 😄

  • All
  • Featured
  • Updates
  • Hackathons
  • Blogs
  • Videos

The World of AI

What better way to get better with AI than experimentation? Here are my trials, experiments and videos where I am just having fun with AI. 🙃

Beatbox

Beatboxing has always fascinated me since my school days. A couple of seniors beatbox battling got me hooked on ever since. Occasionally, I drop a beat or two.

Professional Career

Sumary

Akash James

Highly skilled AI Architect with 3+ years of experience designing and developing high-performance intelligent video analytics pipelines. With possession of rapid prototyping skills, neural network design, and GPU acceleration, I can bring down the development time and produce polished develiverables.

  • Bangalore, India.
  • +91 8971358197
  • akash.james2096@gmail.com/li>

Education

Bachelors of Engineering in Computer Science

2014 - 2018

REVA Institute of Technology and Management

My college journey included me experimenting and chiseling my skills in software development, Internet of Things and Computer Vision. Some of my achievements are as follows -

  • Awarded Brand Ambassador of REVA - 2018
  • Awarded Best Cultural Performer - 2018
  • Won REVA Hack for 3 consecutive years - 2018/17/16
  • 7th place in Smart India Hackathon Finals - 2017
  • 4th place in Rajastah IT Day Hackathon - 2017
  • Won Abhigyaan Hackathon - 2017
  • Founded GLUG REVA - 2016
  • Technical Lead of REVA CSE - 2015
  • Top 30 in Intel Innovation Festival - 2015

Professional Experience

AI Architect

2020 - Present

Integration Wizards Solutions, India

  • Design and construct robust, high-performance pipelines for IW's flagship platform, IRIS.
  • Future-proofing IRIS by building several different modules and integrating them into the singular platform.
  • Develop Intelligent Video Analytics pipelines using NVIDIA Deepstream and TensorRT.
  • Design and train high-performance neural networks with fp16/int8 quantization.
  • Leverage High-performance computing in algorithms to optimise and fully utilise GPUs
  • Optimise IRIS to work on embedded devices such as the NVIDIA Jetson
  • R&D with new technology, constantly improvising and improving IRIS

IoT and ML Engineer

2018 - 2020

Integration Wizards Solutions, India

  • Develop Computer Vision applications for Health and Safety and Retail use-cases.
  • Train classifiers and object detectors such as ResNet18, YOLOv3, CenterNet, etc.
  • Built Kafka pipeline for high-performance inference on multiple GPUs
  • Trained and deployed Fire Detection model using CenterNet and Kafka
  • Built Face Recognition pipeline using FaceNet
  • Develop Flask APIs for services to consume inference results.
  • Accelerate Neural Network performance by leveraging TensorRT.