top of page

Projects

I have worked on numerous projects ranging from Training, Testing, and Deploying ML/DL models as well as Backend Development. These are a few academic projects that I am most proud of:

Network Intrusion Detection System

Python | Flask | Machine Learning

  • Explored and Cleaned the NSL-KDD dataset which has logs of approx. 150,000 network connections. 

  • Applied various pre-processing techniques such as Feature Scaling, Encoding, Sampling, and Feature Selection.

  • Trained several machine learning models such as Support Vector Classifier, Naive Bayes, Random Forest, Decision tree on the pre-processed data.

Screenshot-2021-06-04-at-1-40-28-AM.png

Software Defect Prediction Tool

Python | Flask | Machine Learning

  • Built a web application to help developers identify defects based on the existing software metrics.

  • Performed a comprehensive data analysis to determine the relevancy of each attribute.

  • Trained Decision Tree, Random Forest, and Deep Neural Network models to perform a comparative analysis.

  • Deployed the trained model using pyFlask and used Heroku PaaS to host the application.

68747470733a2f2f692e6962622e636f2f48714c565942712f53637265656e73686f742d323032312d30352d32

Military Waste Segregation System

Python | Deep Learning

• Suggested several architectures that use pre-trained models(and their weights) to classify different types of military-generated waste and segregate them into biodegradable, non-biodegradable, recyclable, and non-recyclable waste.

 

• Trained the CNN models using VGG-16, VGG-19, InceptionV3, and Xception model weights.

Screenshot 2021-05-01 at 9.17.43 PM.png

Self Driving Car

Python | Deep Learning

  • Collected the data required for training from Udacity's car simulator by driving through a dedicated path.

  • Analyzed the captured data and performed data balancing, image preprocessing, and splitting into testing and validation sets.

  • Trained a CNN model based on Nvidia’s End-to-End Neural Network Architecture for Self-Driving Cars.

  • Deployed the Model using Flask and Socket.io to perform real-time testing of the model's behaviour on an unseen track.

9.png
  • g
bottom of page