Automated consumer ticket classification with MLOPS

Classifying financial complaints to relevant departments for efficient resolution. Utilizes NLP text classification techniques such as TF-IDF. Deployed with FastAPI and Docker enabling real-time prediction through a web UI.

Traffic Flow Prediction in the city of Paris using LSTM

This project forecasts the live traffic for the next 3 hours in the famous streets of Paris. Additionally, it also provides statistics for the historial traffic data such as distribution of traffic among 2 wheelers and 4 wheelers, most busiest junction etc.

MERCK hackathon : Predicting readmissions to substance abuse treatment (secured 3rd Place)

Collaborated with a multidisciplinary team to successfully develop a classification model to predict early who will be at high risk for relapse and help in providing tailored support to such patients

Podcasts recommender system using sentence transformers

Collected data from itunes podcasts and spotify podcast dataset. Generated embeddings for all the podcasts and calculated the cosine similarities with the embeddings of the input query from the user. Finally, the top podcasts are recommended. Built a streamlit app and deployed the model on AWS EC2.

Crime rate prediction in San Francisco

This is a project that aims to predict crime rates in real-time using machine learning algorithms and data from different sources, including San Francisco government data, weather data, and basketball schedule data. The project also includes a CI/CD pipeline to automate the process

Data Engineering Pipeline using Airflow, AWS S3-glue, Redshift and Quicksight

Built a Data pipeline using s3, glue, redshift and quicksight. Scraped data from allrecipes.com and ingested into S3 datalake using airflow on AWS EC2 instance. Migrated data from S3 to redshift using Glue job and built a dashboard in Quicksight using SQL queries

Customer retention analysis in the travel industry

This project deals with identifying the probability of a customer returning in the next year. In order to optimize the marketing, the company would like to implement an NBO (Next-Best-Offer) system that predicts which of the previous customers have a high probability of returning in the next year.

Price Optimization in retail energy markets

Customer_analytics_to_aid_marketing_campaigns-using-LLM

Revusense is an AI assitant platform that aims to help businesses gather and analyze customer reviews from Google Shopping and generate a comprehensive summary of the reviews filtered by an input query and rating with the help of SOTA LLM models. It provides tools to create advertisement copy based