PROJECTS
Motivation
It all began when I had a glance on a book series in library : “Tell me why?” and “Tell me how?” in class seventh. I was
fascinated reading stuff about evolution, emerging technologies, Artificial Intelligence etc. From my childhood I was
curious to know how children learn, how humans evolved, how our brain was developed and how it functions: the
mechanisms behind it. My aim in life was clear : “Solving real life problems for the benefit of people”. I wanted to
make something which could replicate human brain. The idea of mimicking the brain always fascinated me. Now seeing
reinforcement learning and AI is like a dream come true. Though we in 2019 are far away from making an exact replica
of human brain in terms of functionality and adaptability, we have started heading towards it. I was fascinated to see
Sophia talk and answer questions posed by us during Techfest in IIT Bombay. I studied how it was made, the way it talks
the database it uses, algorithms and methods behind it. It shows we need to go a long way to make human like robots.
Concept of Agents learning from rewards and punishment was enough to motivate me to dive straight into this world of
agents. I started with Andrew NG’s course on Machine Learning in Coursera. Then I did Machine learning a-z : hands on
python and R which helped me learn and develop various models. After this my interest and confidence in the field grew
tremendously which instilled in me the drive to learn more.
I want to pursue my career in applied research and am truly passionate about it. I was fascinated while using
Reinforcement learning to make Autonomous car : seeing my agent play in the environment was awesome . I developed
AI playing games like Doom and Breakout, made Chatbot using deep learning and Natural Language Processing. I
Have also worked on object detection and image processing. Made fake currency detection system, driver’s drowsiness
detection , drinking liquids purity detector in first year of engineering.
Today, I strongly believe in advancements in Robots and technology as whole. Robots can be sent to places where humans can't go, carry on discoveries and research work in difficult to reach and sustain terrains like under-waters, under-grounds and Space. We humans are limited by Physical capabilities but Robots won't be since we can design them to sustain extreme temperatures and Humidity and can give them desired shape, size to carry out certain specific tasks. Thus robots can supplement us and act as exceptional sources to carry out research work, rescue options and other important tasks.
Latest Projects
Fake News and Website detection system
Internship at Scrapshut
Jan 2029-May 2020
Developed a web-app using Angular and Django for Scrapshut, where users can rate and review various websites by providing their link.
• Wrote a scraper that extracts the title, body, and all the links on the page through URL.
• Trained various ML models like LSTM, XGBoost, CNN on three datasets- Kaggle fake news net, Kaggle: getting real about fake news and Kaggle
fake news Prediction.
• Used pre-trained word embeddings of Glove.
• Also trained a passive-aggressive classifier which is an online learning algorithm to incorporate user rated scraped reviews for real-time prediction.
• Achieved good performance metrics above 93 percent and aced the challenge of making the predictions real-time.
-
Followed Scrum methodology of Software development.
FAKE NEWS DETECTION THROUGH DEEP LEARNING APPROACHES
Jan 2019- May 2019
Real time fake news detection by using Natural Language Processing ,web scrapping and Machine
Learning
BOT AUTOMATICALLY CREATING HINDI WIKIPEDIA PAGES USING WIKIDATA
Nov 2019
Used SPARQL and python script to fetch relevant items through Wiki-data and then converted it to Natural Language. The bot is scalable to any language and can create pages on all topics present on Wikidata.
JETPACK JOYRIDE TERMINAL GAME
Jan 2020
Made a terminal game using OOPS concepts in python
SELF DRIVING CAR
Nov - Dec 2018
Used Deep Q-Learning
Jan 2019
Used Deep Natural Language Processing.Trained Seq2Seq model.
Articial Intellingence playing Breakout game
Nov 2018
A3C algorithm used.
Human Emotions Detector and Recogniser
Sep 2018
Classies 6 human emotions using Haar Cascades and Multi-level Perceptron classication su-
pervised learning model.
Real time Driver Drowsiness and Alcohol consumption detection
Feb 2018 - April 2018
Integrated with Rpi, alcohol sensor and cell phone camera using Viola Jones algorithm. Pro-
grammed in Python.
Quick Draw Dec 2018
Inspired from Google's Quick Draw Project. Integrated with Rpi, alcohol sensor, cell phone
camera. OpenCV, dlib, and Python, Viola Jones algorithm used.
Face Recogniser
March 2019
Facenets and triplet loss function to compare images of different person used.
Articial Intellingence Playing Doom Game
Oct 2018
Built using Deep Convolutional Q-Learning model with Eligibility Trace(n-Step Q-Learning)
June 2018 - July 2018
Using softmax regression
Milk Adultration detector
Aug 2017 - Nov 2017
Mentor : Dr. Debanjan Das, Asst. Professor, ECE, IIIT-NR
Milk sample poured in Parallel Plate Capacitor, prediction made based permittivity.
Street automation and electronic eye
Sep 2016 - Dec 2016
Used LDR, IC LM358 :operational amplier congured in voltage comparator
Movie Booking Application
Sep 2016 - April 2017
Made Using Java NetBeans and MySQL.
May 2018 - June 2018
Object Moment tracking using countour detection and ML