Coding Enthusiast. Deeply intrigued by Data Structures & Algorithms. Mathematics is love. Machine Learning is Life. Gregarious mostly, love meeting new people and exchanging thoughts. Open to technologies and trying out new things.
Often trusts data more than models. Not always. Often prefers Latex over Word. Not always. Often prefers Prezi over Power-Point. Not always.
Taking live sessions on Data Mining and webinars on Artificial Intelligence for 200+ M. Tech students part of BITS’ Work Integrated Learning Program..
Apr. 2019 - PresentWorking for Oracle Digital Assistant team as Data Science and Backend Developer. Studied the latest state of the art research in NLP and developed Named Entity Recognizers for identifying Persons, Locations and rganizations from sentences using Transfer Learning from Google's Bidirectional Encoded Representation from Transformers (BERT) model achieving 99.1% accuracy.
Jul. 2019 - PresentWorked for Oracle Digital Assistant team as Data Science and Backend Developer. Developed a framework for extracting Date, Time, Duration, Set and Email entities from utterances. Designed and established the Kafka pipeline for supporting messaging including devising of Scale out capabilities on Oracle Cloud Infrastructure. Also, worked on developing Kubernetes Cluster monitoring capabilities for Kafka, Elasticsearch and Zookeeper.
Jul. 2018 - Jul. 2019Working for Innovation Center Network (ICN) as a Machine Learning (ML) and Full Stack Developer. Built a model evaluation platform from scratch. Developing Deep Learning and Machine Learning models catering to Natural Language Processing.
Apr. 2017 - May. 2018Worked for Fashion Management Solutions (FMS). Implemented customer enhancements in Sales & Distribution (SD) and Materials Management (MM) modules of SAP ERP System for customers like Adidas.
Sep. 2016 - Mar. 2017Pursued Work Integrated Masters in Technology fully sponsored by SAP. Had weekend classes on various subjects in the SAP campus by professors from BITS. Ranked 1st after two semesters with a CGPA of 9.65. Dissertation Project: Alzheimer's Detection using Neural Networks for 3D Image Classification.
Aug. 2016 - May. 2018Completed my undergrad from IEM affiliated to West Bengal University of Technology. Obtained a CGPA of 9.22 at the end of the degree. Major project was entitled "Data Science in HealthCare".
Aug 2012 - Jul. 2016Worked on Clustering of regions using Time Series Analysis based on similar disease trends. The work was a part of Integrated Disease Survelliance Program, Government of India.
Sep. 2011 - Jun. 2012Worked on developing a hybrid mobile app enabling users to reserve parking slots at various parking venues.
Jun. 2015 - July. 2015Machine Learning
Python
Data Mining
R
Feature Selection
Git
Pattern Recognition
Linux/Unix
Computer Vision
C/C++
Medical/Healthcare Analytics
Latex
Natural Language Processing
HTML/CSS/Javascript
In this research, we studied a residual connections based 3D convolutional neural network for Alzheimer’s Disease Classification proposed in 2016. We improved on its accuracy using an Inception architecture based neural network. We demonstrated the performance of the approach and compared it with the existing approach for classification of Alzheimer’s disease using 3D structural MRI brain scans.
Worked on weekly incidence data of 16 diseases for more than 100 regions in Dehradun. Obtained clusters of regions based on similarity patterns which were subsequently visualized using maps. Also, worked in predicting outbreaks of the diseases and specific regions contributing maximum to the same.
Work submitted to Journal of Indian Society of Remote Sensing, Springer.
Classified Questions from UIUC's CogComp QC Dataset into 6 coarse and 50 fine classes using Named Entity Recognition, Part of Speech Tagging, Lemmas, Syntactic Dependency Relations and Orthography as features. Obtained near State of the Art results with an accuracy of 87.8% over the Coarse classes and 80.7% over the Fine classes using Linear SVM.
Studied 61 research papers to analyze data science research in Health Care with a special focus on India. Highlighted the patient life cycle and identified the role of major entities at various stages of the lifecycle. Identified areas in Indian health system where data science can play a major role. Also, made a comparison between thing happening in the US versus India.
Performed a comparitive evaluation of classification algorithms on Health Care Datasets. Built models for 13 datasets using SVM, Naive Bayes', Classification Trees, Random Forests and Gradient Boosting. Performance was evaluted using Accuracy, Precision, Sensitivity and F1 Score. p-tests were used to study statistical significance.
Studied Hill Climbing(HC) as a heuristic search technique. Variations of HC based on randomness, size of neigborhood and direction of search were also studied. Achieved 1.7% better classification accuracy compared to Genetic Algorithm on 20 datasets. Accepted at Natural Computing for Unsupervised Learning, Springer.
Studied the growth and strategical shift of SAP over the past 8 years using 35000+ tweets. Performed Word Cloud, Hashtag and Sentiment analysis on the tweets. Work submitted to MECS Journal.
© 2019 Aman Kedia
