Android is my favorite smartphone OS. Till now I have developed and published several apps on the play store
Skillsthat I have collected over the span of years
Front End is where I get to use my creative end. Weapons of choice are VueJS and React
PHP was my primary choice during my bachelors but then Node.js stole the crown with its many features
MySQL, PostgreSQL, MongoDB help me secure my data. CAP Theorem has made sure that I don't keep favorites
Got my hands dirty with Image Classification, Text Clustering, Object Detection, etc using good ol' python
Almost killed my Raspberry Pi and my Arduino over a 100 times trying to create different things
Languages I know:
Web Technologies I use:
I manage my Data with:
My favorite set of Tools:
“Any fool can write code that a computer can understand. Good programmers write code that humans can understand”
- Martin Fowler
0Years of experience
0+Lines of Code
0+Cups of Coffee
My ProjectsFew of my strongest contenders
Developed a blend of 2048 and Classic game Snake using plain Android SDK, i.e. without a game engine. The game is published on the play store
Snake is an evergreen classic. I still remember playing it on my old Nokia phone. 2048 on the other hand is a modern classic which I adore. I decided to make something different and came up with this. The snake has a head and tail and body is the numbers it eats througout. The numbers spawned on the screen as food are either a 2, a 4 or an 8. Whatever you eat travels all the way back to its tail and then settles there or combines with a similar number. This snake can bite itself, but biting would then reduce the size of the snake and the number on the body part which was bitten travels all the way back to the tail. The challenge is to keep on eating while keeping the snake short by eating the numbers and to get that 2048.
The game is developed solely on Android SDK. There is no engine involved. The grid here is simply TextViews which are set every 600ms to show the movement of the snake. Since the gridview is linear in nature, the snake movement isn't based on graph math (x & y) but an offset based which depends on the list size.
A web app with a dataset of 370,000 English words to find anagrams and words which can be formed for you to cheat in Scrabble.
It's not everyday that you are playing Scrabble and get dealt a bad set of letters which make you, "Hold my Kombucha! I can write a code for that". Anagram Solver is a web tool made in React coupled along with a API developed in Node to quickly find all the anagrams or words you can form with a given set of a letters. It uses an advance data structure which stores around 370,000 English words for you to go through.
Developed a system which helps detecting forest fires and also predicting the next possible spread of the fire with prediction algorithms
Developed a working prototype for a forest fire detection system using DTH11 and TPM36 sensors and an ESP8266 Node MCU. The main idea behind was to receive a consistent data about the status of the individual sensor nodes and to process this data for finding the next possible area of the spread of fire.
The ESP Node MCU transmitted the data to the server running prediction and simulation. The prediction was based on two algorithms, Boundary Detection and Lane Detection. The next predicted areas were then showcased in the simulations, both in Unity3D as 3D simulation and a Java based lightweight console application with a 2D grid.
According to the recent Amazon fires, 3500 sq mi of area was burnt creating a lot of loss. If this software was deployed, each mile would have 4 TMP36 sensors. 240 sensors are deployed within a range of 60 miles, thus total sensors in the area would be 57600. Each sensor costs about $1.5, thus total of $86400. For this same area, 11520 Node MCU’s are needed, total cost is $23040. Thus, the total cost of sensors, MCU’s and other extra costing would sum up to $120000. Using the containment algorithm, if the fire was a slow spread, of the total 3500 sq mi, 2100 sq mi (9/15th) area could be saved from burning. Whereas, if the spread was at a faster rate, 1633.333 sq mi (7/15th) area would be saved. According to the latest reports, the government issued $22.2 million to combat this situation. Using this technique, for the fast fire $10.3 million and for slow spreads, $13.3 million would be saved.
Here is a video of it in action -
Created an image classifier using Convolutional Neural Networks which detects whether a person is suffering from an eye cataract or melanoma
Eye Cataracts are one of the most common vision defects faced by people above the age of 60. It is very easy to get these fixed with a simple surgery, however it is a shame that cataracts go unnoticed due to lack of medical expertise. Hence, in order to eliminate this, the project aims to provide an easy and free of cost checkup for cataracts in the iris.
An image classifier was developed which is trained with around 500+ labeled images of cataract induced iris and normal iris. The accuracy of the trained model is ~90%, thus making is ideal for medical checks. The model made use of Convolutional Neural Networks which are the industry standard for image classification and recognition. Other models were also created using (Support Vector Machine) SVM and dataset alteration using Histogram Oriented Gradient. The result however concluded that the model trained using Convolutional Neural Networks without Histogram Oriented Gradient filtering on the dataset, proved to be the most accurate. A similar model was also created with images of Melanoma instead of cataract.
Developed an Android app which is a simple to use watchlist for movies and shows for quickly noting down a movie or a show to watch it later.
Being a cinephile, developing this app was more for me than for anyone else. I always get suggestions for movies and I forget them by the end of the day. This app was made to make sure I don't miss out on those gems.
The app receives all its data using the OMDb API in the JSON format. The basic engine of the app is pretty much a few REST calls. I do however emphazised a lot on the UI to make sure that the content delivery is concise and pretty yet elegant. The app is available for download and the source code is also up for grabs.
Created an elegant Android app to help memorize different words along with their pronouciation and meaning for the GRE exam
I thought of making this while I was preparing for my GRE and I did end up making it. Learning a lot of words is definitely a big task and nothing them down is one too. Hence in order to tackle this, I came up with a lucid and elegant UI which helps in learning new words and revising the ones learned before. The app is on the play store and the source is also up in a repo for you to explore.