Sumant Sharma

1. Mined association rules between entities to carry out Market Basket Analysis.
2. Scaled-up the tool by migrating the relational database to Hadoop using Apache Sqoop and sharded the transactional data.
3. Identified the frequent item-sets using Apriori and Parallel Frequency Pattern growth algorithm.
____________________________________________
Platform : Java, R
Framework : Hadoop
...more

1. Trained Support Vector Machine regression model on each of the 1115 drug stores’ data.
2. Predicted 6 weeks of daily sales with an overall accuracy of 94.96%.
____________________________________________
Package: Pentaho timeseriesForecasting API for Weka.
Data Size: 1,017,209 observations, 15 predictors.

Conducted survival analysis using the Cox Proportional Hazards Model to determine that wait time to transplantation is not a predictor of post-transplant recurrence for Hepatocellular Carcinoma.
____________________________________________
Package: PROC PHREG / Kaplan–Meier curve.
Data Size: 172 patients, 21 predictors.

1. Created a logistic response model to predict which customers will respond to a second wave of direct mailings.
2. Captured 95% of the responders by mailing 50% of the customers
____________________________________________
Package: scikit-learn
Data Size: 10,000 customers, 10 predictors.

Predicted V(D)J recombination and mutation burden on cytoskeletal genes using Hadoop.
____________________________________________
Package: Genomic analysis framework ADAM built on Apache Spark.
Data Size: 100 GB Raw sequence file.

Built Random Forest model to predict 5 different ways in which the people workout, with an accuracy of 97%.
____________________________________________
Package: caret, randomForest
Data Size: 19,622 observations, 159 predictors.

1. Developed an interactive visualization tool for healthcare administration in ShinyR –
2. View and filter information based on doctor, hospital, and time.
3. Compare other metrics like wait time, patient load, feedback, rating, revenue, etc.

1. Evaluated their business model, identified the technical requirements and implemented a solution within their budget.
2. Developed RESTful Web Services in the backend for the mobile application for APICON 2014.
3. Worked in close collaboration with the Android and iOS developers to integrate with other applications and enable the push notification service.
4. Learning drug patterns in...more

1. Enhanced and maintained interactive desktop applications for rendering seismic data.
2. Developed interactive toolkits (based on geophysical algorithms) for seismic data analysis and investigation.
3. Migrated functionalities from legacy applications to current JAVA based geophysical data processing applications.
4. Improved performance during the steps associated with interactive model...more

1. Worked on the development, maintenance and testing projects for a global oilfield services major.
2. Agile project execution and management across the distributed teams using SCRUM.
3. Gathered, analysed and documented requirements using UMLs (IBM Rational Software Architect) and System Appreciation Documents.
4. Architectured and designed solutions for suite of GeoSupport applications.
5....more

1. An in-depth analysis of the existing power consumption patterns of wireless sensors.
2. Implemented a wireless sensor network with protocol for power saving using Texas Instrument's eZ430-RF2500 development tool and SimpliciTI™ wireless communication protocol .
3. Demonstrated the ultra-low power consumption of the system by putting in standby mode and waking up in less than 1 μs.

Submitted a project proposal to Motorola Scholar Program - FAER (Foundation for Advancement of Education and Research) whose objective was to develop an inexpensive autonomous guidance system for small vehicle like ATV (All Terrain Vehicle) to perform almost all agricultural processes like fruit picking, soil sampling, row navigation, variable inputs for site specific farming. It also helped...more