To predict delay and severity

Used a CNN for this

A solution to enable upselling and Cross selling

A solution for customer churn

Secured Cloud Communication model with Quantum based SSL and Identity Authentication

Conceptualization, Optimization and Summarization of search results- An overview on Topical web crawlers

A Genetic algorithm based Crypt- Steganography technique for message hiding In Jpeg images

Designed an INCIDENT CRITICALITY AND DELAY PREDICTION SYSTEM for a large pharmaceutical company. This is a NLP driven project that used LDA, WORD2VEC and TFIDF for feature engineering and a stacked ensemble of regressors for the prediction of delay. The project is deployed as a Web API, designed in python- Flask-Django framework.

Currently is associated in designing an AI assistant that prevents the false positives that may happen in OFAC sanctions with SWIFT messaging Monitoring system of a Leading Multinational Bank.

Designed pattern recognizing machine learning models to understand customer churn behavior . Documentation of the processes to enable future analysts to reference Time series analysis to find out the product transitions for various customer cohorts over time. To find suitable opportunities for upselling and cross-selling , and launch of various products. Identifying trending topics on Smart City tag from Twitter API and arriving at a sentiment scores.Done using LDA .

Worked on A FRAME WORK FOR DRIVER FATIGE DETECTION IN VEHICLES USING DEEP LEARNING which used a video stream as one of the inputs along with the vehicle sensor data captured from a camera focused on the driver and the IR sensor from the vehicle .A CNN is used to find the Drowsiness or Fatigue of the driver. A white paper has been published on the company database on this.

Worked for a E-commerce giant for predicting DISCOUNT SEEKERS FROM THE OVERALL CUSTOMER BASE for a given time frame. The company give discounts for all first time users , to encourage them for more purchases. The average revenue per user is calculated for all users from their first purchase for next one year window and also a fuzzy matching algorithm is used on the shipping address details .A Random forest classifier is used to predict .

The role comprised of developing, updating and maintaining a health Software package Health Watch in VB 6.0 with back end MySQL. Was in charge of organizing training programs for the Data processing Supervisors and Data Entry operators on the proposed front end of the software for periodical population update surveys of the Adopted community Key roles included the analysis of the collected data using various statistical tools and pass on the end results to the higher management for research or remedial measures. This included identifying population where a certain type of inoculation was required or identifying a potential area with a likelihood of some epidemic break out.