SABAREESH MAMIDIPAKA
MACHINE LEARNING & DATA SCIENCE
CONTACT
adcnsz@r.postjobfree.com
Santa Clara, CA
linkedin.com/in/sabareeshmamidipaka
github.com/sabareesh169
PROFILE
Love to work in a fast-paced, hands-on
and collaborative environment to work
on real-world datasets and help
organizations make better data-driven
business decisions.
EDUCATION
2019
PURDUE UNIVERSITY
Master of Science in Statistics &
Mechanical Engineering
2017
INDIAN INSTITUTE OF TECHNOLOGY MADRAS
Bachelor of Science in Mechanical
Engineering
TECHNICAL SKILLS
Database: MySQL, PostgreSQL
Data Visualization / Dashboards:
Tableau, matplotlib, seaborn
Data Analysis: Numpy, Pandas,
Scikit-learn
Programming: Python, R, C,
Matlab, SAS
Deep Learning: TensorFlow
Big Data: PySpark
Recommender systems
Convolutional Neural Networks
Natural Language Processing
Bayesian Methods
COURSES
Data Mining, Deep Learning, Machine
Learning, Uncertainty Quantification,
Applied Regression Analysis, Statistics
and Probability
PROJECTS
Developed a custom algorithm – Neural Forest
The algorithm combines the principles of Random Forests, DNNs and bagging, increasing the accuracy of predictions by over 20% for missing value datasets when compared to DNNs.
Built ML pipeline to predict target audience
Applied Logistic Regression, Decision Trees and Random Forests to predict the users most likely to click on an Ad.
Dynamic variation of threshold implemented to suit the business problem at hand. Obtained an accuracy of 98% on the test set. Linear regression model to predict house prices
Data cleaning, feature engineering using box cox transformation and various diagnostics like t-tests, VIF and cook’s distances were performed to make the model robust. Obtained R2 value of 0.95 Walmart – Multisite time series sales forecast
Wrote a Python script to build time series, ensembles and hybrid models for each department in each store to forecast sales.
Achieved an MSE of less than 3000 and finish in top 50 on the Kaggle leaderboard.
Physics informed Deep Learning to predict velocity
Implemented a Deep Learning model (Python-TensorFlow) which predicts the flow field from the location of particles in the flow images. Solves the unsupervised problem of particle matching and supervised problem of field prediction.
The model combines the knowledge of Mechanical Engineering and Deep Learning to obtain the same accuracy as the existing methods by using only 10% of the data.
Predicting seniority level from resume data
Performed data cleaning and feature engineering on unstructured data to extract skills, major and highest degree.
Data insights and visualization was performed and an ensemble model was built to classify the applicants into 8 groups of seniority. Recommender System for hardware tools
Built a recommender system to recommend new products in the catalogue based on the current items in the cart using a hybrid algorithm which combines the semantic and behaviour data. EXPERIENCE
2018 - 2019
Teaching Assistant, Purdue Object-oriented programing
Gave course lectures every week to over 50 students teaching programing in Python. Conducted office hours to help students in their assignments, project works and explain concepts. 2015 - 2016
Intern, Caterpillar R&D
Modified the existing software written in MATLAB for beam analysis to account for the Fiber-reinforced plastic beams.