Roja Immanni
addrma@r.postjobfree.com
www.linkedin.com/in/rojaimmanni
EDUCATION
University of San Francisco, San Francisco, CA (July 2019 - June 2020 (expected)) Master of Science in Data Science
Coursework: Machine Learning, Deep Learning, Recommendation Systems, Algorithms, Time Series Analysis, Linear Regression Analysis, A/B Testing, Data Ethics, Product Analytics Indian Institute of Technology, Madras, India (Aug 2011 - May 2015) Bachelor of Technology in Engineering
PROFESSIONAL EXPERIENCE
University of California, San Francisco
Deep Learning Researcher Intern (Oct 2019 - Present)
● Created efficient deep learning architectures for image problems on medical datasets
Down-scaled the models by over ~50 times with similar performance while saving on training costs and time
● Trained and evaluated the state of the art CNN models on medical datasets with and without transfer learning
● Explored different training strategies for Neural networks on Imagenet data Urban Ladder (furniture e-commerce company based out of Bangalore, India) Associate Product Manager (Jan 2017 - May 2018)
● Reduced logistics costs due to cancellations by 35%. Enforced 2-step confirmation process before processing for orders with a high probability of cancellation.
Used a logistic regression model with order/item details, user demographics, past purchase behavior, activities
● Proposed a retail store layout by identifying highly associated products using the Association Rule mining.
Resulted in improving walk-in cart size by 1.3 times compared to that of online
● Improved on-time deliveries to 98% from 83%. Developed an ETA prediction model for each step in the delivery life-cycle, identifying and raising flags for any possible delays
● As the company moved to an omnichannel distribution strategy, analyzed and built a Revenue attribution model for multiple sales channels based on the touchpoints
● Setup a centralized data warehouse and automated MIS and curated low-level business reports. Configured cohort analysis, retention, churn and reactivated user metrics as per business goals Graduate Trainee (managed a team of 4 engineers and 2 QA testers) (June 2015 - Dec 2016)
● Reduced the overall call and email volume by 40% while improving customer experience by designing self-service features to handle cancellations, refunds, and status tracking related requests
● Automated the logistics process for the complex wardrobe orders from supply to installation
Reduced the delivery timeline by 3.5 days and led to a 15% cost reduction in the overall process
● Revamped the ‘Last-Mile Delivery App’ with the objective of enhancing customer experience & achieving operational efficiency at the final step; improved NPS score by 4 basis points and reduced delivery time to 1/4th PROJECTS
● Human Activity Detection: Predicted human activity with 83% accuracy based on the time-series data from different sensors and smart devices using a deep learning model with SparkML and H20 on AWS EMR
● Implemented core Machine Learning algorithms from scratch with comparable performance compared to scikit-learn packages, including Ridge, lasso, Naive-Bayes, Decision Trees, Random Forests, Boosting, Kmeans LANGUAGES AND TOOLS
● Languages: Python (Pandas, Pytorch, Scikit-learn, Matplotlib, Pyplot, ggplot2), SQL, R
● Tools & Framework: Spark, Google Analytics, SiSense/Tableau, Google Bigquery