Rutuja Moharil
Graduate Student
(University of Pennsylvania)
Hi , I am Rutuja a second year Master's student at Upenn . Please take a moment to explore my site, where you’ll find details on my background, experience, skills and more.
I look forward to meeting and connecting with people.You can reach me via email at rutuja@seas.upenn.edu
About Me
A deep learning and signal processing enthusiast, skilled in machine learning, biomedical imaging,data analysis,dynamics and control . Seeking challenging and exciting full time opportunities from May 2020 onward.
Here's a link to my resume .
When I am not working on networks and not thinking of loss functions and images , I love watching and exploring new dance routines. I am a trained dancer and I love dancing .
I like travelling and I am a self proclaimed photographer :-P
My Experience
April 2019 - Present
Research Assistant @ Penn Advanced cardiovascular imaging Lab
Developing the pipeline for automated ventricular short axis cardiac MRI multi-class segmentation using U-net architecture.
Using segmentation contours calculate ejection fraction as a metric for detecting myocardial ischemia.
Extend segmentation results to perform correlation studies for genome and phenome-association (Keras).
May 2019 - August 2019
Machine Learning intern ,HeartVista Inc
Employed deep learning-based image reconstruction ,similarity search for MRI images and built an image search database for efficient and faster search, in a peer reviewed and agile workflow.
Optimized image embedding dimensionality using convolutional neural networks and performed unsupervised clustering analysis to draw clinical insights to help physicians provide streamlined patient treatment.
Graduate Trainee Engineer, Siemens Ltd
July 2017- July 2018
Worked in the Nagpur Metro Project for Rail electrification business unit.
Actively involved in the material control and Supply chain management in Nagpur metro project.
Developed and presented requirements management and material control modules for Nagpur Metro Project.
Research Intern, IIT Bombay, India
May 2016 - July 2016
An Indian government funded initiative aimed at developing and promoting an open source software (Scilab) as an alternative to MATLAB .
Developed the open source Control system toolbox framework for numerical computation .
Developed the code for key function libraries ,for analyzing,modelling of systems in Scilab.
Developed and designed a GUI in Scilab to help students analyse, model ,plot and understand single input single output type of systems with a single click of a button
Education
University of Pennsylvania
August 2018 - May 2020
Master of Science in Electrical & Systems engineering.
GPA : 3.433/4
Coursework : Deep learning,Machine learning, Signal Processing, Biomedical Image Analysis,Optimization,Control Theory and Linear Systems.
Visvesvaraya National Institute of Technology, India
August 2013 - May 2017
Bachelor of Technology in Electrical and Electronics engineering
GPA : 8.22/10
Coursework : Control system, Sample data and digital control ,Electrical Machines, Calculus and probability
Narayana Vidyalayam
April 2009 - May 2011
Physics ,Chemistry ,Mathematics stream.
Marks obtained : 94.6%
Narayana Vidyalayam
May 2011 - April 2013
Physics ,Chemistry, Mathematics,Social studies,Sanskrit.
Marks obtained : 10/10 GPA
My Projects
Cardiac MRI segmentation
Data Wrangling : 1) Image data collection from proprietary software(SuiteHeart) to .mat structures.
2) Automated extraction of metadata (raw image , contours ) from .mat structure to png file format.Image segmentation : U-nets based Right ventricle (RV) and Left ventricle (LV) segmentation from short axis cardiac MRI .
Co-relation and data analysis : Extend performance results from segmentation to draw correlation studies to identify risk factors and biomarkers .
Unsupervised representation learning with Triplet loss
Employed autoencoder network architecture using triplet loss and reconstruction loss to extract deep representations and image clusters from unlabeled image data (Tensorflow).
Data visualization using PCA and TSNE of raw pixel data and discriminative cluster representations (Tensorboard).
Performed analysis for optimal encoder embedding dimension and visual cluster representation .
Deep learning based vehicle speed estimation
Employed deep learning based speed estimation of an autonomous car in real time from mounted dashboard video stream using FlowNet (Keras).
Employed dense optical flow technique (Farnerback method) as a discernible metric for speed calculation between pair of successive frames.
Adaptive Equalization
Developed an adaptive equalizer with a training input to reduce Intersymbol interference
Developed adaptive equalization algorithm for the case of blind equalization so as to learn the input sequence using gradient descent and least means square algortithms
Performance comparison of Breast Cancer prediction
Worked on data pre-processing involving region of interest detection and extraction of training patches (Keras).
Performance comparison analysis of deep learning convolutional neural network architectures like ResNet ,DenseNet to predict cancerous cells in lymph node tissue scans.
Achieved test accuracy of 96% using ensemble of ResNet and Densenet and an AUC of 98.3 %.
Finding Kinematic constraints on chain of unicycle robots
Employed leader-follower configuration (formation control) of multi-unicycle robots.
Achieved the leader follower formation using control strategy from a follower's frame of reference.
Implementation of d-phi approach on the kinematic constraints of leader and follower by simulating various trajectories using MATLAB and achieved stabilization by minimizing error between desired and simulated trajectories in MATLAB.
Publication
Presented research paper at 11th IEEE conference on Intelligent systems and Control at Coimbatore, India.
Abstract :
This paper presents the development and features of a Scilab based user-friendly CSIT (Control Systems Interactive Tool). It is an Open Source Graphical User Interface (GUI) tool that contains basic functions to learn and analyze Control Systems at undergraduate and graduate levels. Using this tool, one can compare the system plots and parameters of different systems in time and frequency domains. The tool also supports a “know concept” button, represented in the form of “?” symbol adjacent to the function to display the concept related to the function. A student is not required to learn any programming language to use this tool.
Link : https://ieeexplore.ieee.org/abstract/document/7855953
For more examples of my work follow my github link
Skills ,Leadership and Awards
Programming Languages and Frameworks
Languages :
Python
C
MATLAB
Libraries:
Tensorflow
Keras
Pytorch
Pandas/scikit-learn/scipy
OpenCV
Git/Bitbucket
Scholarships and awards
Received Scholarship for AI in cardiology seminar by Mayo Clinic.
Society of Women engineers Profession development committee member in Upenn .
Recipient of the Ex-servicemen welfare scholarship for exemplary performance in 12th board exams.
Recipient of INSPIRE(Government of India) scholarship for securing a rank in the top 1% in 12th board exams.
Hobbies
Dance :
I have been grooving to literally almost any type of songs /beats since I was a child .
Won several group dance and solo dance categories. (Bollywood ,Hip Hop ,Semi classical)
Travelling : Big fan of exploring new places and travelling on a budget ! Lived in more than 8 cities in span of 15 years !!
Table tennis : Although just a novice , I enjoy playing table tennis and secured second position in Intra college torunaments in undergrad.
Professional Development
‘Intro to Self-driving cars ‘, Udacity.
‘Control of Mobile robots ‘, Georgia Tech by Prof. Magnus Egerstadt.
‘Introduction to CS101x ‘, IIT Bombay, Prof D.B Pathak.
'Machine Learning',Coursera ,Andrew Ng.