SaaSifying Dealer Marketplace (D2C) for CARS24 🏎️ | SDE III - Backend @CARS24 | Ex - Probo, AyuHealth, Harness, Clear |

github

medium

youtube

Navneet Das  • Peerlist

A scalable/maintainable/extensible/modular/object oriented implementation of Cricket Batting Scoreboard in Java

A cricket scorecard that shows the score for a team along with score of each player.

Given the number of players in each team, the number of overs, their batting order as input, and overs data (ball by ball) with the runs scored on that ball (could be wide, no ball or a wicket as well).

The project prints individual scores, number of balls faced, number of 4s, number of 6s for all the players from the batting side at the end of every over. It also print total score, total wickets.

Essentially, it keeps a track of all the players, strike changes (at the end of the over or after taking singles or 3s) and maintains their scores, also keeps track of extra bowls that are being bowled (like wides or no balls).

It also prints which team won the match at the end and type of win (wickets/runs/draw).

These are the bare minimum features which are there in the project. The plan is to add
some more features, like maintaining bowlers record (total overs bowled, runs conceded,
wickets taken, maiden overs, dot balls, economy, etc.). Total team extras, batsman
strike rates, etc. can be added too. But these are "good to have" features, have currently
tried to complete the bare minimum only.

The code is readable, maintainable and object oriented. It is written in a modular and extensible way, to add new features if needed.

Cricket Scoreboard

Contributed with Java as an android developer to DecisiO, Hack-A-Bit which served as a daily life suggester based upon user’s past choices.

Aim: To provide a personal suggester in form of an android app. This app initially has capabilities of suggesting 5 things, namely Food, Cloth, Activity to do, Music and Place to visit.

Tech Stack: Python, Android, Machine Learning, Flask API, and Heroku (for deployment).

Work flow: 
1. The features for the above mentioned 5 queries were decided.
2. Features include Salary Range, Age, Per Month Expenses, Gender, etc.
3. A total of 102 features were written.
4. With the help of Data Generation techniques, dataset of 5000 units was made.
5. Various types of Machine Learning algorithms were used.

Following accuracies were obtained :-
1. Random Forest Classifier: 70.4 %
2. Decision Tree Analysis:  65.2 %
3. Logistic Regression: 40 %
4. Support Vector Machines (SVM): 44.5 %
5. K Nearest Neighbours: 71.4 %

Separate models were made for each query, i.e. 5 different models were trained for 5 queries. Each of them was hosted on Heroku. Then, using Post requests and Flask API, the model weights and data were updated continuously. The android app contains questions for the user. He/She will be asked for a frequency for questions to be asked regularly at the start of the app. After that, he will have 2 options either to ask a question or to answer a question.

DecisiO

A NodeJS implementation of FPL API fetch to list all players of a team

https://fantasy.premierleague.com/api/bootstrap-static/

Further plan is to extend the script to take two inputs - position (element_type) and team to print all the players that play in a particular position for a team and all the statistics for each of them.

Also, plan is to add 2 more inputs that take a particular stat name and stat threshold and returns a further filtered list for all players that have a value for a stat that is greater than the threshold.

FetchFPLAPI

Objective -
Presently AyuHealth's customer app has very minimal interaction with the customers, so we wanted to pitch an idea, which will make it more interactive with the customers.
We planned to integrate their customer app with Google Fit and Apple HealthKit, so that we could get data related to customer's daily activity and health related information.

User Stories we covered -
- Integrate with Google Fit and Apple HealthKit
- Design interactive dashboard
- Design smart notifications
- Redirect users to book appointment according to health data (Via notifications)
- Redirect users to videos/blogs (Via dashboard)
- Redirect users to community (Via dashboard)

I worked on the Apple HealthKit integration for the ReactNative iOS application, and a bit of research on the notifications and background service to fetch daily health data continuously and take actions based on that.

HealthAssist (AyuHealth's Internal Hackathon Project)

This is a Java based open source android project for HackFest 2017 called "hubISM", which served as a solution to all the daily life problems of IIT(ISM) students. The app was made to be usable even in offline mode. The app follows an open-source approach for further development in addition to what the team did itself. The app consists of these features -

1. Detailed map of IIT(ISM) Dhanbad including all the departments, lecture halls, event locations, labs, and places of interest.
2. Detailed timetable for all the courses. (currently for only first year of 2016-2020 batch)
3. Contacts containing photos and contact details of all the faculty and other important persons.
4. MIS hyperlink to directly access features such as attendance, grade sheets (currently opens in external browser window)
5. Academic calendar includes all the vacations, exam dates, and other important dates.
6. Clubs - this feature consists of a brief introduction of every presently active club in IIT(ISM) Dhanbad with their logos. It also includes posts made on their facebook page available at a single place, known as "Club Feed".

hubISM

A tool to test the PHP integration of InstaMojo APIs using Apache2, MariaDB, PHP 7.1 and Webhook for processing payment for an item.

InstaMojoTest

A log watching server implementation similar to tail -f in Java.

Further plan is to leverage WebSockets for client < - > server communications and Queues for logs.

This project implements a log watching solution (similar to the tail -f command in UNIX). However, in this case, the log file is hosted on a remote machine (same machine as your server code). The log file is in append-only mode.

The project implements the following:

A server side program to monitor the given log file which is capable of streaming updates that happen in it. This runs on the same machine as the log file. The server is implemented in Java. It also includes functionality to fetch the last 10 lines of the log file.

Further plan is to include a web based client (accessible via URL like localhost/log) that prints the updates in the log file as and when they happen and 'NOT' upon page refresh. The page should be loaded once and it should keep getting updated in real-time. The user sees the last 10 lines in the file when he lands on the page.

Project Constraints -

- The server should push updates to the clients as we have to be as real time as possible.
- Log file may be of several GBs, server should be optimised for retrieving the last 10 lines.
- The server should not retransmit the entire file every time. It should only send the updates.
- The server should be able to handle multiple clients at the same time.
- The web page should not stay in loading state post the first load and it should not reload thereafter as well.
- We should not use any off-the-shelf external libraries or tools to read the file or provide tail-like functionalities.

Log Watcher

A C++ implementation of GPU accelerated image processing using Python OpenCV for conversion of any coloured image into its greyscale version.

rgbTogrey

Trained and predicted the electricity market data using NNTool in MATLAB & the best working structure of an electrical micro-grid considering solar and other energies as parameters. (Under - Dr. Nitai Pal, IIT (ISM) Dhanbad)

Solar energy electricity market data prediction and Power-gr

An extensibly/modularly/cleanly coded system design for SplitWise

Functionalities:

1. User can split spend with another user
- Based on percentage
- Absolute values
- Equally

2. User can split spend with a group
3. User can exclude few people from the group while adding expense
4. User can simplify debts - A owes money to B, B owes money to C. If B chooses to simplify debts, then A will pay to C directly

Future plan: User should be able to export their monthly summary of expenses and dues

The code is extensible, modular and clean. Choice of data structure was important for extensibility of the code. Took some time to think through it. Started working from the first requirement and modelled other requirements on top of it.