Help & Supportcareers@gravityer.com
Director, Data Scientist at Mastercard
AI is analyzing your overall score…
Identifying your key strengths…
Evaluating your skill match against the job requirements…
Assessing your cultural and operational fit
Director, Data Scientist at Mastercard and Graduate from The University of Illinois at Chicago with a Master's Degree in Electrical and Computer Engineering with a major in Deep Learning, Computer Vision, and Processor Architecture.
University of Illinois Chicago
Master’s Degree, Electrical and Computer Engineering
January 1, 2015 – January 1, 2017
Visvesvaraya Technological University
Bachelor’s Degree, Electrical, Electronics and Communications Engineering
January 1, 2011 – January 1, 2015
Mastercard
Director Data Scientist
March 1, 2025 – Present
Mastercard
Lead Data Scientist
March 1, 2022 – March 1, 2025
Mastercard
Senior Data Scientist
March 1, 2020 – March 1, 2022
Mastercard
Data Scientist II
September 1, 2018 – March 1, 2020
Launchpad.AI
Data Science Associate
May 1, 2018 – September 1, 2018
San Francisco Bay Area
Fellowship.AI
Machine Learning Fellow
January 1, 2018 – May 1, 2018
Greater New York City Area
Waec LLC
Machine Vision/ Machine Learning Engineering Intern
October 1, 2017 – December 1, 2017
Chicago, Illinois
TrueMedicines
Machine Learning Intern
August 1, 2017 – December 1, 2017
Chicago, Illinois
Digital Jalebi
Summer Intern
May 1, 2014 – July 1, 2014
Bengaluru, Karnataka, India
'Cannabis Advisor' development using DialogFlow
October 1, 2017 – Present
Designed an AI-powered chatbot using Google's DialogFlow to provide the best of Cannabis solutions for a given condition. The agent also uses Firebase as its backend for the fulfillment. The agent is currently deployed on Google Assistant awaiting approval for publishing.
Vision model for the Waec's UGV using SSD Mobilenet and TensorFlow
October 1, 2017 – December 1, 2017
Built an Object Detection model which would contribute to the Vision pipeline prototype of Waec's UGV. The model was re-trained on Google Cloud Platform's ML Compute Engine and will be used in tandem with vanilla computer vision algorithms as an extension to read the counter on pedestrian cross walk signal. The model was designed using TensorFlow and Python
Transfer Learning on the STL-10 dataset using Inception and Convolutional Autoencoders
August 1, 2017 – Present
Used Stanford's STL-10 dataset and Inception V3 model to bottleneck the labeled examples and performed transfer learning by constructing a simple Softmax classifier. Also, the STL-10's huge unabeled dataset was used to design a Convolutional Autoencoders (CAE) to extract features from the unlabeled dataset. This training is done on the Google Cloud Platform. Once the CAE is trained, the labeled dataset is bottlenecked until the encoder's last stage (ReLU) and the dataset is now classified using a simple Softmax classifier.
Image Painting using TensorFlow
June 1, 2017 – Present
A simple Artificial Neural Network is fed an image after pre-processing it. The deep neural network is programmed to work as a regression model. The inputs are the 2-D coordinates and the label is a 3-channel gray scale value at that coordinate pair. While training the model, the prediction of the model is displayed as an image. This gives the feel of the model "painting" the image it has seen. The model yields better results with higher training epochs.
Transfer Learning on Inception Model using TensorFlow
June 1, 2017 – Present
Google's Inception Model is known to be one of the smartest Deep CNN models trained on the ImageNet dataset. In this project, work was carried out to bottleneck all the layers but the penultimate two layers of the model with a personal dataset comprising of faces of my people. Once the bottlenecks are generated, the model is now a simple Softmax regression to classify the faces between classes. The next step in the project is to feed faces/frames from a LIVE video feed to classify the faces as friendly or non-friendly faces and provide access.
Gesture Recognition using Leap Motion controller and LSTMs using Python, and Keras Deep Learning library with TensorFlow with GPU (CUDA) back end
March 1, 2017 – May 1, 2017
The Leap Motion controller is used to train Gestures and a Time Series dataset of 60 frames is generated for every valid gesture and free gestures each. The dataset to used to then train a Recurrent Neural Network (LSTMs) model in classifying the Gestures. The model is saved and is then loaded to the main source code predict the trained gestures after it is performed to the Leap Motion controller. The recognized Gesture can be used to perform any suitable task as demanded by the application, i.e. Swipe Right or Swipe Left, etc. Python, and Keras with TensorFlow back end are used for the project along with the power of GPU for training the RNN.
Gesture Segmentation using Skeleton Tracking and Machine Learning
November 1, 2016 – Present
Involved the bifurcation of Skeleton data as gestures to control a window (Ball, in this case) or non-gestures using KINECT and training of the data using Machine Learning algorithms such as SVMs, Decision Trees, and Backpropagation Neural Networks to make the bifurcation more efficient and accurate.
Classification of MNIST Images into numbers using Logistic Regression and Neural Networks
May 1, 2016 – Present
The standard MNIST examples are used to train a deep learning neural network comprising of 2 hidden layers. The experiment was performed both on MATLAB and Google's Tensorflow. The speed of computation is more efficient in Tensorflow due to its capability of handling the network as graphs. The error rate on Test data in Tensorflow is very low as well.
Performance Analysis of ARM's big.LITTLE architecture against homogeneous processors by varying different Scheduling Policies
April 1, 2016 – Present
Simulated and compared an octa-core ARM big.LITTLE architecture comprising of 4 big (A15) cores and 4 LITTLE (A7) cores against 2 homogeneous configurations of all 8 cores of A15 and 8 cores of A7s by measuring the Power consumed, core consumption efficiency using Snipersim. The scheduling policies of the processor were also re-configured and the Power Consumption was compared for different configurations
Real Time Conversion of Sign Language into Audio and Text output to assist Hearing and/or Speech impaired- IEEE EPICS
December 1, 2014 – May 1, 2015
This project aimed at developing an interpreter between the speech impaired and an ASL illiterate by making use of Computer Vision in Leap Motion Controller and a custom designed Arduino board for playing the word and displaying text. Won the Second Best Project for the Dept. of ECE Open House 2015, RNSIT
Machine Learning and Neural Networks
Coursera
June 24, 2026 – Present
AMCAT Certified Content Developer - Electronics and Semiconductor Engineering
Aspiring Minds
June 24, 2026 – Present
AMCAT Certified Design Engineer - Electronics and Semiconductors
Aspiring Minds
June 24, 2026 – Present
AMCAT Certified Engineering Trainee - Electronics and Semiconductor Engineering
Aspiring Minds
June 24, 2026 – Present
Digital Image and Video Processing
Coursera
June 24, 2026 – Present
Cultural Fit Analysis
The candidate's diverse project portfolio, ranging from academic research to industry applications, demonstrates adaptability and a broad interest in various ML domains. The experience at Mastercard, a large enterprise, suggests an ability to work within structured environments. The personal projects indicate a strong drive for continuous learning and innovation, which aligns well with a dynamic technical culture.
Soft Skills & Operational Fit
The candidate's project descriptions indicate a proactive and problem-solving approach. The progression through various Data Scientist roles at Mastercard suggests strong operational fit and ability to contribute to R&D initiatives. The IEEE EPICS project highlights teamwork and practical application skills.