Resume

 Auto Subtitled Video Generator

• This project is an automatic speech recognition application that takes a YouTube video link or a video file as input to generate a video with subtitles and a transcription as .txt, .vtt, .srt file.

• The application performs 2 tasks:

  • Detects the language, transcribes the input video in its original language.

  • Detects the language, translates it into English and then transcribes.

• Generated a transcription of the video using the OpenAI's Whisper model.

• Generated a subtitled version of the input video using ffmpeg.

• Built a user interface using the Streamlit library, hosted on HuggingFace Spaces.

​

 Customer Churn Predictor

• This project is a customer churn prediction system in a fictional telecommunication use case.

• Trained and evaluated various machine learning models such as Logistic Regression, SVC, Random Forest Classifier, Decision Tree Classifier, XGBoost Classifier, LightGBM Classifier.

• Performed feature selection using Recursive Feature Elimination technique and tuned the hyperparameters for the best performing model, the Logistic Regression model.

• The model predicts whether the customer is likely to churn or is happy with the services, with an accuracy score of 81.1% and an F1 score of 80.6%.

• Saved the best model using joblib library and used it to make predictions  in the web application.

• Built an interactive web application using the Streamlit library and deployed it on Heroku.

​

News Summarizer

• Built functions to scrape the news from Google News using Beautiful Soup library.

• Categorized the news as trending news, favorite topics or you can search a topic of your choice.

• Fetched news, news posters and summarized the news using Newspaper3k library.

• Built the web app using Streamlit and deployed it on Streamlit Cloud.

​

TV Series Recommender

• This project is a tv series recommender system.

• Created a dataset of TV series by scraping the IMDb website using BeautifulSoup library and saved the dataframe to a csv file.

• Built a K-Nearest Neighbors model using Euclidean distance metric to make recommendations based on user’s preferences.

• Built an interactive web application using the Streamlit library.

• Retrieved and displayed the trailers of recommended TV series dynamically, using Youtube Data API v3.

• Deployed the web application on Heroku.