RISHAV SINGH

Meticulous with data-driven and technology-focused approach. Communicates clearly with stakeholders and builds consensus around well-founded models. Talented in writing applications and reformulating models. Understanding of statistical, algebraic and other analytical techniques. Highly organized, motivated and diligent with background in Tensorflow,Scikit-learn,Keras,NLTK,Spacy,.

POC - Movie Recommendation System

Technology Used: Python,Streamlit,Scikit-Learn,TMDb API,GIT

This Model recommends top 5 movies with posters to user based on the input movie by the user.

• It was an End to End machine learning project.
• The dataset contained 5000 movies of different genres and languages and was obtained from Kaggle.
• After Performing Explorartory Data Analysis on the dataset, Used Lemmatization on the Tags to reduce words and afterwards performed Vectorization on the words.
• Used Cosine Similarity for finding similarity between movies.
• Used Streamlit for building the web app for the Recommender System and Displayed Visual Posters of the movies through TMDb API.
• Finally Deployed the model through Heroku.

Cotton Disease Prediction Using Deep Learning

• Technology Used: Python,Scikit-Learn,InceptionV3,Tensorflow,Keras
• Cotton disease Is Primally Caused By These Category: Fungal, Bacterial, Insects.
• The Dataset contained 2000 training and 100 test images of cotton plants obtained from Kaggle.
• The Model Uses Transfer Learning To Predict If a Provided Image Of Leaf Or Plant is Infected.
• Used InceptionV3 From Keras as it is Computationally Less Expensive And Contains 48 Layers .
• The Output layers were connected to the Pre-trained Input Layers,With four nodes to classify the type of cotton disease and used Softmax activation.
• This Obtained new neural network architecture was trained with the weights freezed for the Transfer Learning layer
• The Model Gives an Accuracy Close To 94% On the provided Images of Cotton Leaf Or Plant.

Heart Failure Prediction

• Technology Used: Python,Scikit-Learn
• This Was A Classification Problem That Whether A Given Patient Provided With His Different Details Such As His/Her Age,Blood Pressure,Smoking Habits Have A Effect On His Heart Failure.
• Using Different Explorartory Data Analysis, I found The different Correlations Between The Independent And The Dependent Variable.
• Applying GridSearchCV and testing it on different SkLearn Classifier.Got XGBoost Classifier to perform the best with an Accuracy Of 94%.

Object Detection Using TENSORFLOW

• The Code Trains and Leverage Custom Object Detection Model using the Tensorflow Object Detection API On Real Time
• Used OpenCV for reading the video and images and to draw Contour Lines for Object detection.
• Gesture detection: The Algorithm Detects Multiple Gesture's like different Hand signs like Thumbs up,Thumbs down,Thank you etc.
• The Algorithm Is Also Trained On Sentiment Analysis Which it does On Real Time.

3

Tabular Playground Series - Jan 2022 (Kaggle)

• Top 100 In Over 1300 Teams (Solo Contributor)

WiDS Datathon 2022 (Kaggle)

• Top 50 In Over 250 Teams (Solo Contributor)

Deep Learning Specialization