aky91

Akash Yadav

I'm good at Java and Image Processing.
I like the concept of machine learning.
Giving the ability of thinking and decision making to a computer is very fascinating.

Problem-solving is what drives me. I like to solve problems by using technical knowledge. I like to work on things that have the potential to bring benefit the society.
There are many things that already exist. We just need someone to come up and ask "Hey, what if we combine these things? Will I get something different? something helpful?"

Top Skills

Top Projects:
1> Genetic Image Evolution.
It was my own idea. The inspiration came from Daniel Shiffman book Nature of Code. It was made in Processing to study Genetic Evolution on Images.
I was amazed by seeing 'survival of the fittest' in action.

I learned that the fittest entities must have some advantage over other entities in passing genes to the next generation. Also, setting the mutation rate is critical. Too low, and mutation will have no effect. A little too high and it won't let the population evolve.
Also, we need to have an exponential fitness curve to accelerate evolution.

2> Single Pixel Camera
What if I told you, it is possible to capture images with just a single pixel with some really clever optimizations and maths(Compressed Sensing). This technology will drastically reduce the cost of the security camera and an infrared camera. My team build this during my internship at DRDO.

I learned a lot about work ethics. I realised that coding is the very last part of a project. First, we need to put a lot of effort into the research work for better understanding of the problem and deciding the solution approach.

3> Brain Controlled Home Automation
We used the Brain-Computer Interface Device(BCID) to read the signals of the brain and all you have to do is just think about turning the light off. It was done by predicting thoughts by analyzing brain signals using machine learning.

4> Edugent
Rural India is facing teacher's crisis where the post of over 1 million teachers is vacant. For sure there exist NGOs but they don't have access to proper teachers for rural areas. Keeping this in mind, we are developing a platform which minimizes the gap between the NGOs and the rural areas in terms of education factors such as teachers, books, etc. The teachers who are interested in volunteering in the educational field in rural areas can register on our app. NGOs will have the information of all the volunteers regarding a particular subject and standard. According to the requirement of the class, the teacher can be selected from the list available to the NGO through our app. Based on the performance and talent of students, volunteers can provide recommendations for institutions for their higher studies.

5> GeoARgraphy
An augmented reality geography guessing game made in Unity.
Students have a lot of books to read. Reading from books is a slow and boring experience. Humans generally only remember 10% of what they read. This makes it harder to retain information.With the increasing burden to study, students are constantly stressed.
According to a study, 49% of the school students are under great stress.

What we did was to make an interactive augmented reality geography game that asks users to guess the location of a country on an augmented reality Earth model. People tend to overestimate their knowledge of geography. This is here to actually put their knowledge to the test and learn something in the process. Our app was intended for young learners and for anyone who is interested to learn while having fun at the same time.

Projects

GeoARgraphy

An augmented reality geography guessing game made in Unity.Sketch, Android, Unity, Augmented Reality, Vuforia

PathMarkAR

? An augmented reality indoor navigation app made in Unity and ARCoreARCore, ARKit, Unity, Unity3d, C#

Skills

Python
Java
JavaScript
TensorFlow
Chatbot Development

Experience

  • Defence Research and Development Organisation - Summer Intern
    May 2018 - July 2018

    Mentor: Dr Sudhabrata Majumdar (Scientist E DRDO)

    Project: Single Pixel Camera using Compressed Sensing

    Project description visible light Single Pixel camera was made using Compressed Sensing method.

    Pre-requisites: Image Processing, Amplifier circuit, Optics, Compressed Sensing, MATLAB

    Set up: the object, a PC for generation of random black and white pattern and further computing, a visible light photodiode, lens, amplifier circuit. arduino.

    The object is kept in front of the random pattern. Using lens the light intensity is read by the photodiode, which further sends it to the PC. The program then reconstructs the original image by using Compressed Sensing algorithm called Basic Pursuit.

    Importance: if we have an image of 2500 pixels then using only 500 patters we can reconstruct the image good enough to determine its shape and size.

    Applications: low cost and power cameras, IR Imaging, Hyperspectral imaging, Defence(border monitoring system)

    I learned a lot about work ethics.
    I realised that coding is the very last part of a project.
    Before that, a lot of research is required for a better understanding of the problem statement and planning.

  • Citi - Programming Analyst
    May 2019 - July 2019
    • Designed and Developed Protegrity Template for App360. Tokenized apps data using Protegrity's DTaaS.
    • Worked in a team to create an automatized survey to collect information on various apps in the SST(Securities Services Technology) Department