This is a collection of my personal projects that I work on in my free time. Hope you like them.
TensorNet
A high-level deep learning library built on top of PyTorch.
- The library contains some advanced concepts like GradCAM and LR Finder into the package so that they can be used via a simple function call.
- Extensive use of code modularization and OOP has been done with proper docstrings for each function in order to maintain a clean and understandable codebase.
- Tools: Python, PyTorch.
- PyPI URL: PyPI
- GitHub URL: TensorNet
Topic Based Image Captioning
An automatic image caption generation system built using Deep Learning.
- Developed a model which uses Latent Dirichlet Allocation (LDA) to extract topics from the image captions.
- Developed a caption generation model using LSTMs which takes the image features from a pre-trained InceptionV3 network and the topics from the LDA-model as input.
- Made the caption generation model using merge model architecture.
- Tools: Python, Tensoflow-Keras, NLTK, OpenCV-Python, MSCOCO-2017 Dataset.
- Services: Google Cloud
- GitHub URL: Topic-Based-Image-Captioning
Stock Bridge
A real-time online stock market simulator built using Django.
Website URL: https://stock-bridge.herokuapp.com
- Built the entire user-company transaction system from scratch.
- Developed an automated system for fluctuating stock prices.
- Developed a scheduler mechanism for automating user transactions.
- Extensive usage of django signals, model managers and custom querysets.
- Developed a Bank Model to issue loans and deduct interests from users.
- Tools: Python, Django, Django REST Framework, Bootstrap v4, chart.js
- Services: Heroku, sendgrid
- GitHub URL: Stock-Bridge
Code Warrior
An online judge platform built using Django.
Website URL: https://nitmz.pythonanywhere.com/
- Built the entire compilation, execution and submission evaluation module from scratch.
- Designed the platform to support languages: C, C++, and Python.
- Constructed a tiebreaker mechanism which uses user submission execution time for ranking users with the same score in the leaderboard.
- Tools: Python, Django, Bootstrap
- Services: Amazon Web Services, PythonAnywhere, sendgrid.
- GitHub URL: Code-Warrior
Kart
An E-commerce website built using Django.
Website URL: https://shan-kart.herokuapp.com
- Built the backend entirely on Django. Used jQuery in some places to make the website asynchronous.
- Used signals, custom model managers, and custom querysets extensively to keep most of the code logic within the models and to make the communication between the linked models effective.
- In case of a server error, setup sendgrid to send a detailed error report to website admins.
- Built the functionality to sell digital items by storing them in AWS S3 Storage.
- Render the order summary as a PDF and send it to the user after a successful transaction.
- Tools: Python, Django, Bootstrap, jQuery, Ajax, jsrender, chart.js
- Services: stripe, mailchimp, Amazon Web Services, heroku, sendgrid
- GitHub URL: Kart
Autoranking Amazon Reviews
Ranking the reviews on Amazon according to their helpfulness score.
- The problem was modeled as a regression problem. The performance was evaluated by using the coefficient of determination and rank correlation.
- Predictions were made based on various categories of features of the review text, and other metadata associated with the review, with the purpose of generating a rank for a given list of reviews.
- Tools: Python, Numpy, Pandas, textblob, scikit-learn
- GitHub URL: Autoranking-Amazon-Reviews
Morphosis, NIT Mizoram
Android App for the annual technical fest of NIT Mizoram.
Google Play Store URL: Morphosis
- Contains all the information of various events to be conducted during the technical fest.
- Live notification updates.
- Contains a game called Scooby Dooby Doo, where participants have to answer various logical questions. The app gives live leaderboard updates.
- Tools: Java, Android Studio, Firebase
- GitHub URL: Morphosis-2k17-Android