| (XXX) XXX-XXXX | XXXX@XXXX.XXX | Portfolio: | |
| | Tempe, AZ | SKILLS Programming: Python, SQL, Java, HTML&CSS, JavaScript
Libraries and Frameworks: Numpy, Pandas, SciKit-Learn & TensorFlow, Keras, OpenCV, Django
Visualization libraries: Matplotlib, Seaborn, Folium, D3.js
Machine Learning: Regression techniques, ensemble methods (random forests, gradient and XGBoosting), CNN architectures,
time series forecasting, computer vision techniques (object detection and Pose estimation)
Other Tools: MATLAB, AWS Sagemaker, Tableau, Jupyter, Power BI, G-Cloud, Git & version control PROFESSIONAL EXPERIENCE Data Science Open Lab, Tempe, AZ: Research Assistant
Aug 2018 - Sep 2019
• Collaborated with ASU School of Earth and Space Exploration to automate the process of Crater Detection.
• Built an initial baseline model by segmenting images of the lunar surface, manually labeling sub-regions that contained
craters, and training a binary classifier to classify each sub-region as either containing or not containing a crater.
• Built a second modeling pipeline using pre-labeled data, image-segmentation and object detection algorithms like Mask R-
CNN and YOLO. The final model obtained an F1 score of 88.7 for crater detection.
Wipro Limited, Bengaluru, India: Project Engineer
Aug 2016 - July 2018
• Designed and analyzed A/B tests in collaboration with product managers and engineers. Provided data support for product
decisions and new feature launches that improved signups by over 12%.
• Rewrote our automated query and complaint tracker from Perl into Python/Django to enable greater scalability, speeding
up webpage loading by nearly 20%.
• Designed and maintained dynamic reports for key decision-makers. Reports included P&L analyses for different divisions of
Wipro, and on the utilization of our consultants at all client sites, allowing executives to make informed decisions on billing
and allocation.
ACHIEVEMENTS AND ACTIVITIES • 3rd runner up in Edureka’s online Techathon competition for making best regression model.
• 2nd place in Pixy-Challenge in Hack Arizona Hackathon for making a person tracker robot using a pixy camera. PROJECTS Detect an Isolated Zoom Action: FOX ML Hackathon
• Implemented InceptionV3 CNN model on video footage of basketball games, to detect instances when the camera was
focused on a single player. The top layers of the model were trained using 500,000 labeled frames from basketball games,
using AWS SageMaker. The final model gave a test accuracy of over 90% and was selected to be presented to the Judges
from the FOX Technology team.
Location Recommendation system for opening a car rental business
• Built a model to help small companies (e.g. restaurants) choose the best location to open their business. Gathered a
dataset by scraping Wikipedia and using the Foursquare API. Recommended locations based on similarity of existing
neighborhood businesses to the user's business, and on the likelihood that the user's business will obtain a high
customer rating relative to the average for that neighborhood. Reconstructing Noisy Images from the Fashion-MNIST Dataset using Auto-Encoders • Developed denoising autoencoders to reconstruct noisy images from the Fashion-MNIST dataset. Fine-tuned the
autoencoder by varying hidden layer nodes, learning rate and noise level in the input images. Also developed stacked sparse
autoencoders and performed greedy layer wise training with different levels of noise to classify the images. Achieved a final
classification accuracy of 78% on the fine-tuned model.
EDUCATION Master’s in Computer Engineering Aug 2018 - May 2020
Arizona State University - Tempe, Arizona
Bachelors in Electronics and Communications Engineering
Aug 2012 - June 2016
PESIT - Bengaluru, India