Description

A unified platform for industry intelligence and decision making. Vitrina’s mission is mapping and connecting the cross-border video entertainment value-chain so as to enable safe, secure and speedy transactions. It is B2B SaaS platform. It Enable media companies to close licensing transaction faster by providing a related suite of application.

Roles & Responsibilities

• • I am working as a backend developer. My most of the task are related to creating backend REST API’s.
  • Writing backend queries using SQLAlchemy to fetch or insert the data from the database. Reprogramming existing database to improve functionality.
  • Writing effective and scalable python codes. Debugging application to ensure low- latency and high availability.
  • Test cases using Pytest python module. And also creating services using AWS lambda. Assessing and prioritizing

Description

Zinia is the platform for the organizations or people who don't know machine learning but want to use machine learning technologies to grow their business. The purpose of this platform is to democratize machine learning and bring value machine learning in the hands of data analysts/business analyst. We created a very simple and user-friendly UI. We used many AWS services.

Roles & Responsibilities

○ In this project my job is to develop the backend of this web app and databases i used is MongoDB, file storage system is S3 bucket, hosting platform is AWS instance.

○ And the backend framework is Django. So I created REST API’s in Django and backends function to do all the processing.

○ In this project we divided the whole application in small microservices and we created a docker container of every microservice

○ Worked on full lifecycle of the project including requirement analysis, development, testing and defect fixes

○ Implemented docker to containerize REST API as micro services and they are deployed in Kubernetes cluster

● Description

This project was for organizations like BPO’s, we were developing the solution particularly for work from home scenarios. In this project the most important and

challenging task was deployment of the application which does actual processing. So we figured out that docker and the EXE is the best way to deploy the application on a user's machine. So we divided our application in two parts one is EXE and the other one is docker. And for installing docker on users machine, we combined our exe with docker exe and made an msi installer.

Roles & Responsibilities

• • Developed a backend application using Flask serving a REST API and Dockerize the whole solution.
  • This project includes Face detection, Face recognition, mobile detection, Unknown detection, multiple person detection using OpenCV and Deep Learning algorithms.
  • Identifying the non-compliance activity by accessing the webcam of the user machine and deploying the solution in docker and integrating the solution with the UI.
  • Sending non-compliance activity to the rabbitmq first and from rabbitmq consuming those non-compliance frames.
  • Also implemented CI/CD pipeline in git-lab, which includes automatic docker image creation and code testing after that pushing the image to the organization docker hub and exploring the different possible solutions for deployment of the application.

● Description

It was a small module of a very large application, I created it separately because we wanted to implement the same case in two different applications. The difference in two cases was in the first one we were using a webcam and possibility of multiple faces was very low and in the second case we were using security cameras. Our aim is to detect in which direction a person is looking.

Roles & Responsibilities

• • Detecting the gaze of the eye using dlib, OpenCV and Deep Learning algorithm.

The first step of this project is to detect faces using YOLO. After using 68 landmarks in dlib detection eyes and at last detecting the position of the pupil. By detecting the position of the pupil, the model is able to identify the coordinates in the frame in which a person is looking

● Description

Parking Detection system monitors the actual occupancy of a parking lot, provides its managers with valuable information, and navigates drivers all the way to an empty parking spot. This was also part of a smart mall application. Input for this application was parking lot images. I trained a TensorFlow model to detect whether parking lot space is vacant or not.

Roles & Responsibilities

• • Detecting a parking lot is vacant or occupied using OpenCV and Deep Learning algorithms.
  • Input in this project is Frames from live video from cameras installed in the parking lot.
  • First step is to detect whether the space is a parking lot or not. If it is a parking lot, finding out it is occupied or vacant using classification technique. For this project I take some assumptions like camera mount on top so that it can capture most of the view clearly.

● Description

I did multiple projects of machine learning for learning or practicing. This was one of the projects I did to predict the amount of insurance claims. This is the one of the unpaid competitions from Kaggle, so i used the dataset from Kaggle. I used different machine learning algorithms and chooses the one which gave the best result and that is the Random forest algorithm.

Roles & Responsibilities

• • In this project my aim is to predict the amount of the insurance claim using various features. I used the dataset from UCI Machine learning library and performed EDA on that dataset.
  • I tried with many machine learning algorithms, but the best output is from the Random Forest algorithm. After that I did model serialization using joblib python library.