Analyzing sound data and trying to predict machine performance and health using audio processing techniques and Machine Learning / Deep Learning , which is deployed in AWS Lambda.

Outcome:

• Collected sound data from ships and performed Data Analysis by generating features like mfcc, zero crossing score ,spectral etc,.
• Applied Machine Learning / Deep Learning Algorithms like CNN,SVM ,Auto Encoders to validate and finalize the models with 90?curacy in Anomaly Detection.
• Built docker image and deployed to AWS lambda service and added S3 trigger to invoke it.
• Created an end to end pipeline to UI by inserting all the predicted data to InfluxDB, MySQL database.

Generating recommendations for users based on their video history in a virtual conference using Data Analysis and Machine Learning Techniques.

Outcome:

• Collected description and title from the past conference data ,performed data preprocessing
• Built item-item similarity model ,generated embeddings to do transfer learning on new conference data.
• Computed user-user similarity indexes.by analyzing user history.
• Recommended top10 videos based all computed similarity scores.
• Created a docker image and Deployed it as REST API using FastAPI module which is similar to Flask.
• Programmed in Python, Machine Learning and inserted data into MySQL using SQL.

Categorizing promotional data into multiple labels using NLP.

-->Predicted multiple labels for a single promotion(for a single record we need to predict one or more labels).
--> Worked closely with client based on Miami(flexoffers.com) and figured out data pre processing techniques to clean the data.
--> Applied Natural Language Processing(NLP) techniques and pre trained embeddings
--> Merged several tables in MySQL to gather data
--> Trained Multi-label classifier using LSTM with hyper parameter tuning like pytorch,tensorflow and Scikit-Learn.
--> Created a pipe line to process thousands of data with multiprocessing techniques which reduced processing time gradually.
--> Built a docker image on python and configured it as a cronjob, which processes new records daily.
--> Started Exploring Flask

Categorizing different products based on their name and description using Data Analysis and Machine Learning/Deep Learning.

--> Worked on categorization of product information into 5000 different categories.
--> Extracted the labelled categories from millions of products from MySQL Database.
--> Worked closely with client based on Miami (flexoffers.com) and figured out data pre processing techniques to clean the data.
--> Applied Data Analysis and Semi supervised learning approach to train a model using Machine Learning/Deep Learning Algorithms which fits best for the labelled products.
--> Generated more labelled data which is used to train LSTM model using pretrained embeddings. with modules like pytorch ,tensorflow and Scikit-Learn.
--> Created a pipe line to process thousands of data with multiprocessing techniques which reduced processing time gradually.
--> Used Team Foundation Server repository to push the code which similar to GitHub.
--> Built a docker image on python and configured it as a cronjob, which processes new records daily.

Web Scraping and generating reports based on the sensitivity of the web pages by applying Machine learning and Data Analysis.

--> Written script to crawl web pages using python libraries like Beautiful Soup, Urllib, NLTK.
--> Classified the data using Natural Language Processing(NLP) and Machine Learning / Deep Learning techniques and stored the data to MySQL DB.
--> Created a Flask to deploy the project as an REST API.
--> Used modules like PyTorch and TensorFlow.

Project invoves detecting the duplicate/similar questions asked in quora.

Identify which questions asked on Quora are duplicates of questions that have already been asked. This could be useful to instantly provide answers to questions that have already been answered. We are tasked with predicting whether a pair of questions are duplicates or not.