Let’ talk a bit about the major projects and courses that I have done in field of cs. I took a computer vision course as an elective in the college where I learned about different types of filters, morphological operations, feature extraction and detection techniques, homography matrices and 3d reconstruction. I am certified for ML by andrew ng on coursera, deep learning on coursera and data science with python. I also have thorough understanding of cloud computing domains including AWS and GCP which I have explored in few of my projects.

Have dedicated enough time to develop basic understanding of different architectural styles per-say monolith architecture, the problems associated with it and micro services architecture. I have few projects that included the use of docker and kubernetes, basically it was a web based application, which I’ll come a little later and then an understanding of the various domains in big data. In particular I have done Hadoop cluster setup and written a element counting algorithm in map reduce paradigm for this purpose.

So now coming on to the projects, I’ll just focus on few relevant projects. First project is dog breed classification API running on GCP, it’s a rest api service which runs on a server using flask in python. In order to accomplish that, I created a GCP envirnoment. On that deep learning model was created using keras flask and docker which was deployed as an end-point to act as a as rest api. I uploaded the container into dockerhub, created a kubernetes instance on the GCP and deployed the dockers on it, so i have multiple servers. Using docker and kubernetes enables low latency, scalability and fault tolerance.

Talking a little bit about dockers, docker is one of the best containerization technology present today. Container basically are like virtual machines but they are different in the sense that they all share same the kernel as the operating system unlike virtual machines. Virtual machines tend to directly connect to the hardware using hypervisor but in containers they interact with os kernel with their independent dependencies and files. I can basically run and deploy any software in docker. Now coming to kubernetes, it’s an orchestration platform for running containers. In kubernetes, we have a master kubelet and other nodes where kubeltes are installed. On thes, docker image is installed and a yaml directive is wrutten to describe the architecture of the kubernete cluster.

Currently also I am working on a project based on object detection and IOT. Basically it’s a cap for blind people. So in this I will have a cap with 6-8 cameras around and it will reconstruct a 3-d world and detect an object, once detected it will measure the distance of the object with the user, and then output an audio file stating that object x is at a distance y in this particular direction. Although this is quite a big project but, we are thinking of patenting this. Till now, what we have accomplished is taking the input of two cameras using ov7673 module and two Arduinos, which we will shift to raspberry pi, and using the video which is generated at 30 fps, we will create a opencv program for 3d reconstruction of the video using kafka and spark, and then do object detection on it. Once the object is detected, we’ll determine the distance of the object, as I’m having two cameras that can easily be possible. I will convert this into a voice file and with direction we’ll send the signal to the audio output to user. So this is the workflow of the project. It’s certainly an ambitious project an d we’ll start it and make a smaller prototype and later our junior batches may carry forward it.