Harshana

Harshana Sridhar

Bio
I'm a machine learning enthusiast who is really passionate about data and obtaining insights from it. A cup of coffee, listening to melody songs and coding late at night is what drives me.

The most complex project I have worked on involved DEEP LEARNING. It was to classify CT scan images of lungs of patients into one of the five different types of Tuberculosis. I was provided with images in NIFTI format which is the format for 3D images. I sliced the 3D images into 2D images, applied the given masks on the middle layers and extracted ROI(region of interest). Then a CNN model was developed and the images with extracted region of interest was given as input for it. It was a 16-Layer model. I achieved an accuracy of 75.6%. This project was developed during a 36 hour hackathon and we were awarded the BEST WOMEN TEAM.

I was responsible for the preprocessing part and it helped me realise the importance of preprocessing and that it is the most crucial step in any project. Before that, I used to concentrate on building a model and henceforth, I started concentrating on pre-processing any data given to me.

I believe that garbage in is garbage out; so any unstructured data must be preprocessed and this is the first step in any process and this is what excites me. Seeing a raw data transform into insights is itself a beautiful process. Though this seems like a small step, this is the root of any great model starting from Google's ranking algorithm to high end facial recognition systems. If you cannot generate suitable data any great idea can collapse into failure and with great data, any mediocre idea can give great results.

I'm a great believer that practical knowledge contributes to actual learning experience. This belief made me take part in numerous hackathons even before the completion of my second year and win Best Women Team and Top Coders in MLH Local Hackday hosted by Github Inc. This year, I was even part of one of the top 4 finalists teams in NATIONAL LEVEL SMART INDIA HACKATHON 2019, which took place in Lovely Professional University, Punjab. We developed an AI based model which clusters colleges according to their features. It was a great learning experience and we achieved great results with our model.

This June, I participated in WMN hackathon organized by Devfolio and I was really surprised to find how wonderfully it was organized with mentor support from industrialists throughout the hackathon.

I am sure that this hackathon will be the right platform to showcase my talent and skills as well as be a great learning experience to improve myself as a developer.

Projects

She-Farer

Get, Set, Travel!HTML, Bootstrap, Vue.js, CSS, JS, Flask, Google Maps API

F.R.I.D.A.Y

A Natural Language Processing based Datascience Coding Assistant for the Dyslexic.Flask, SQL, Artificial Intelligence, Data Science

Skills

Python
Java
SQL
C++
C