Machine learning,Data Science, Algorithm Development Engineer

PRANATHI CHUNDURU

** ******** **, **** ****, CA -94015 LinkedIn: http://www.linkedin.com/in/pranathichunduru

480-***-**** ********@***.*** Github: https://github.com/pranathichunduru Biomedical

Engineering

graduate

with

strong

foundation

and

experience

in

areas

of

machine

learning&

statistical

analysis,

seeking

challenging opportunities

to

explore

new

frontiers

in

the

field

of

Machine

Learning,

Data

Science,

and

Algorithm

development. CORE COMPETENCIES

Machine Learning: Supervised & Unsupervised learning algorithms, Decision Trees, Linear Regression, Logistic Regression, KNN, SVM, Neural Networks (CNN), Principal Component Analysis, Bayesian Statistics, Bag of words (NLP). Related Experience: Design and implementation of statistical / predictive models, Data visualization, Time-Series Analysis, Software Testing (Unit, Regression), FDA (510k, 21 CFR Part 820,821)

Digital Signal & Image Processing: Time –Frequency domain analysis, Computer vision, Pattern Recognition, 2D/3D Image Reconstruction,Sensor DSP SKILLS

Programming Languages: C, C++(with OOP), Python(Numpy,Pandas, Scikit Learn), MATLAB Statistical Tools & Databases: SQL, SAS, JMP, WEKA MySQL, SQL Server Big Data Technologies & Modeling Software: Apache Spark, Tensor Flow SIMULINK, NI LabView Operating Systems & Version Control: Windows, Red-Hat Linux(BASH), OSX Git EXPERIENCE

ST Microelectronics Algorithm Engineer Intern May’2016- Aug’2016 Analog & MEMS Group – Santa Clara, CA.

• Evaluated performance of attitude filter in MEMS sensor based pedestrian dead reckoning system (PDR) from high dynamic human motion data.

• Determined transformation matrix to align gyroscope data (quaternions) simultaneously collected from smartphone and motion capture system(Tele- Immersion lab -UC Berkeley). Implemented scripts to predict missing values and performed statistical analysis to determine orientation error.

• Analyzed walking data using PDR stride length model for estimating step length at various gait patterns to aid in indoor navigation. Used regression algorithms to redesign the stride length model for accurate prediction. Arizona State University Research Assistant Dec’2015- May’2017 Neuromuscular and Human Robotics Laboratory – Tempe, AZ.

• Developed novel rehabilitation system to determine walking symmetry for post-stroke(movement disorder) subjects by extracting step-length data using 3D motion capture system(VICON) and generating virtual visual feedback.

• Used LabVIEW/MATLAB to stream real time data from motion capture using VICON software development kit(SDK). Visualized step-length data results in JMP and Performed statistical analysis using one-way ANOVA, and post-hoc (Tukey’s) on obtained step-length ratio data. (MATLAB). Indian Institute of Sciences Research Fellow May’2014- Aug’2014 Jawaharlal Nehru Centre for Advanced Scientific Research- Bangalore,INDIA

• Conducted behavioral experiments to study thermo-sensory behavior of Drosophila melanogaster under different temperature cycles.

• Collected continuous stream of activity/rest rhythm data using Drosophila Activity Monitor (DAM) and processed binary data using DAM Filescan 102X software (Trikinetics FaasX). Cleaned and organized data into bins to extract activity parameters. Plotted Actogram and Eduction graphs using FaasX and Microsoft Excel tools.Performed statistical analysis to visualize results of experiment to make data driven experimental decision. EDUCATION

Arizona State University (Tempe, AZ) –Master of Science Biomedical Engineering Aug’2015-May’2017 Osmania University (Hyderabad, INDIA) –Bachelor of Engineering Biomedical Engineering Aug’2011-May’2015 MOBILE APPS

Workout motion sensing app (Link to App : Fitzzer -

https://itunes.apple.com/us/app/fitzzer-smart-personal-fitness/id1002919900 )

• Worked with developers to test and debug motion-tracking algorithms. The application uses accelerometer and gyroscope data to detect and track exercises. Implemented MATLAB scripts to identify false positives and reduce true negatives by collecting user data at various activity patterns. ACADEMIC EXPERIENCE

Human Activity Recognition with Smartphone – Kaggle Data Science Participation (Python –Numpy, Scikit learn, Pandas) - Github

• Implemented Machine learning algorithms on human activity dataset containing accelerometer and gyroscope sensor values recorded using smartphone worn on waist. Developed scripts using Python libraries for feature extraction and to fit best classifier model using decision trees and KNN. Used K fold cross validation to verify the accuracy of model and to minimize false negatives. Visualized the results using Matplotlib tools. Predicting trajectory noise for vehicle tracking (Kaggle - Data Science Participation) (Python –Numpy, Scikit learn, Pandas)

• Implemented Machine learning algorithms to keep track of trajectory noise for accurate vehicle prediction in self-driving cars. Computed noise from large dataset containing clean & noisy data and trained supervised learning model using extracted features from data.

• Used k-fold cross validation to verify accuracy of the model. Implemented scripts using Python -Numpy, Scipy libraries. Automated Carotid Intima-Media thickness measurement from Ultrasound Videos (MATLAB, Python) - Github

• Developed image processing and computer vision pipeline for measuring thickness of carotid artery from ultrasound video frames.

• Designed MATLAB –Graphical user interface (GUI) for automatic Region of Interest (ROI) selection, implementation of deformable SNAKE algorithm and determining absolute difference between boundaries (thickness). Deep learning application for computer aided polyp detection system from Colonoscopy videos(MATLAB)

• Implemented Convolutional Neural Network and Random Forest algorithm for classifying colonoscopy images of different subjects for polyp detection. Binned polyp and non-polyp images as 1’s and 0’s and used these binarized images to train the classifier.

• Plotted ROC(receiver operating characteristic) curve to determine precision & sensitivity of the model and verified accuracy using 10-fold CV. PUBLICATIONS

• Neuroscience: Pranathi Chunduru, Gabrielle Maestas, Hyunglae Lee ‘Effect of implicit visual feedback distortion on gait modulations in multiple walking speeds’, in Journal of NeuroEngineering and Rehabilitation (JNER), 2016.

• Conference: Pranathi Chunduru, Seung-Jae Kim, Hyunglae Lee ‘Locomotor adaptation on implicit feedback distortion and split belt treadmill’ in Biomedical Engineering Society (BMES) Phoenix, Arizona, 2017.

Contact this candidate