Remote Asset Discovery Tool (2021)
Go/Golang based tool that is able to monitor local and remote (via SSH tunnel) systems and their networks. It also provides an API and CLI.
Winning solution of Flipkart GRiD 3.0 (Information Security Track) National Hackathon.
Stack: Go, ElasticSearch as a database, Kibana for visualization, VueJS
Khuim: AI Based Virtual Tourist Guide (2020)
Khuim is an all-in-one virtual tourist guide, which is available as an Android/iOS App and a progressive web app. The following features were implemented :
- The user could take a picture of a monument from their phone, it would be sent to the django based server, and then evaluated using a resnet pytorch based model (applied transfer learning with custom data on pretrained resnet model). After detection, information from a central database would return back to the user, with stuff like the monument name, history, visitor-timings etc.
- Flutter based multi-platform application, and the backend was made in django. In the app, you could look at monuments and places to visit around you, book a cab, or get the bus schedule.
- ChatBot in the Help and FAQ section, which used DialogFlow for Natural Language Processing, and could guide the user on multiple types of queries.
- Map section using Google’s Map API, that could display your location, help you navigate to a set monument or restraunt, or evaluate which ones are currently open, and traffic information.
- Plan my trip section, with pre-planned trip packages, suggesting which places to visit in what order, so as to ensure optimum planning.
- Section for Emergency Contact information, One-touch dials for the Local Police, Fire Brigade, or the Child and Women Helpline.
- Feedback section to help identify bugs and improve our app.
Winning solution of an intra-university hackathon.
Stack: PyTorch, Django, Flutter, DialogFlow
Helexia AI Cameras (2019)
- Helixa is our attempt to simplify and disrupt current security and surveillance practices, by developing smart artificial intelligent cameras that won’t require the hefty network, DBMS and manual labor resources to operate and maintain.
- To implement this project, we used two deep learning models on Violence Detection and Object Detection. This helped us achieve features like Anti Theft, Anti Violence, Inventory Management System.
- Further an administrative console and a telegram bot were synchronised to execute the tensorflow and cuDNN OpenCV backend on the raspberry pi, along with a Flutter based cross platform notification application.
Second Runner Up solution of HashHacks 4.0 National Hackathon.
Stack: Flask, Keras, Flutter, Telegram Bot
Interaction system for students and clubs/societies. Every society gets their own admin accounts, and a web portal where they can upload information regarding society events. Students can register for events and ask querys.
Stack: Django, Flutter
Startup Punjab Portal (2019)
Web portal for startups and governments. Government can set up and upload schemes for startups and startups can sign up, see which ones are applicable to them and apply. Seperate interfaces for the government and startups. Other features: Newsfeed for Startups about schemes applicable to their area of interest. Graphs and metrics for the government.