Karishma Agrawal



I'm a Graduate student at Carnegie Mellon University. I am pursuing Masters in Information Systems Management, with concentration in Business Intelligence and Data Analytics. My primary interest is in the field of Machine Learning and Data Anatytics.



Master of Information Systems Management,
with concentration in Business Intelligence and Data Analytics.

Carnegie Mellon University, Pittsburgh
GPA 3.55/4.33


Bachelors of Technology, Computer Engineering
Sardar Vallabhbhai National Institute of Technology,
Surat, India
GPA 9.22/10

MAY 2014


Research and Development Engineering Intern JUNE 2015 - AUGUST 2015
Sony Corporation, Tokyo

I worked on developing solutions based on deep learning technology for the task of sleep detection and human activity detection. Specifically, my solutions were based on convolutional neural network and recurrent neural network. The new algorithms performed much better than the previously employed solutions both in terms of accuracy as well as decrease in latency. The end results would be used in post development phrase. The coding was done in python with the aid of a wrapper based on theano.

Data Intern MAY 2013 - JULY 2013
Impetus InfoTech Pvt. Ltd., Indore, India

I worked, independently, on developing a part of the programmer friendly, scalable Machine Learning Library. This library was used to support and simplify the development of other projects within the company. I used this opportunity to study and gain hands on experience with various Big Data technologies such as Apache Hadoop, Apache Cassandra, Apache Mahout, MapReduce and Weka. The crux of my internship included programing of Machine Learning algorithms such as kNN on MapReduce paradigm and optimizing their performance on the cluster.

Research Intern MAY 2012 - JUNE 2012
Sardar Vallabhbhai National Institute of Technology, Surat, India

I assisted Professor Dipti Rana in her research of temporal apriori algorithm. I implemented a basic version of the apriori algorithm and then worked on improving it using sliding window. I studied the association rules generated from a super market dataset and the variation in their confidence with respect to different time frames

Summer InternMAY 2012 - JUNE 2012
Sardar Vallabhbhai National Institute of Technology, Surat, India

I worked under the guidance of Dr. M. A. Zaveri and carried out a literature survey on Cellular Automata and its applications. I implemented cellular automata algorithm for image processing, encryption and data mining (image classification) using MATLAB.


Multilingual Path RankingMARCH 2015 - MAY 2015
Carnegie Mellon University, Pittsburgh

Agumenting knowledge bases with dependency parsed text has been proved to be effective in creating stronger connectivity in the graph. Specifically, performing random walks over this augmented KB has shown significant improvement in learning relationships and semantic inference rules. In this project, the focus was on incorporation of lexical syntatic labels from multiple languages. The underlying idea was to identify the merits of KB reinforced with Subject-Verb-Object(SVO) triples from multilingual text corpus and test different methods of establishing connection between different languages. The goal was to incorporate text of one language into the KB and see if we could learn information in another language. More details can be found in the project report.

Pump It Up: Data mining for Tanzanian Water CrisisMARCH 2015 - MAY 2015
Carnegie Mellon University, Pittsburgh

Water is critical to a country’s development as it is not only used in agriculture but also for industrial development. Though Tanzania has access to a lot of water, the country still faces the typical dilemmas of many African countries where many areas have no reliable access to water. In a household where money is scarce, families have to often spend several hours each day walking to get water from water pumps. Our motivation for this project is to look at the dataset for water pumps in Tanzania to predict the operating condition of a waterpoint. By finding which water pumps are functional, which need some repairs, and which don't work at all, the Tanzanian Ministry of Water can improve maintenance operations of the water pumps and make sure that clean, potable water is available to communities across Tanzania. Information regarding the methods used to identify the pumps which need repair can be found in the report.

Human Activity RecognitionJANUARY 2015 - MARCH 2015
Carnegie Mellon University, Pittsburgh

Attempting to predict the physical activity of a person, via health meters and other sensors has been a subject of research in the medical research community for a long time. In this project we have been given a small dataset of 4 subjects with their physical attributes such as gender, height weight, BMI and the readings from four accelerometer placed at different parts of their body. The objective is to come up with a reliable generalized predictive system, which would accurately depict the person's physical activity. After initial analysis the major risk identified, amongst others, was to capture the temporal nature of the data. To mitigate this flaw, I worked on generation of new features based on the sensors parameters. This was followed by segmentation of newly generated features into sliding time windows and finally feature extraction and reduction. The analysis concluded with major improvement in prediction power and confidence in change point detection. Further details can be found in the project report.

Data Warehousing for Loan Acceptance and RejectionJANUARY 2015 - MARCH 2015
Carnegie Mellon University, Pittsburgh

Using the loan acceptance and rejection data available at the Lending CLub, my team and I have populated a data warehouse. I carried out the data preprocessing task using R, and designed the data model. We intend to use the warehouse to answer several business questions, which would enhance the user experience by providing them access to the trends in loan statistics. Our primary business goals provide a clear insight regarding the loan acceptance pattern, and various underlying criteria.

Object Recognition using Discriminative Features and Linear ClassifiersDECEMBER 2014
Carnegie Mellon University, Pittsburgh

As a part of my term project, I explored the Machine Learning approaches applied in the real world to classify objects. Amongst the prevailing methods used for object recognition, the deep learning methods are most favoured. Rather than continuing what has become a trend to improve the former methods, I tried to explore the time honoured linear classifiers such as Support Vector Machine(SVM) and Multinomial Logistic Regression(MLR).My team demonstrated how simple transformations can make the data more responsive to these straightforward linear separation rules and produce results comparable to the generic feature based classifiers with complex framework. To further evaluate the relationship between the transformations and performance of the linear classifiers, we carried out various feature engineering experiments on MATLAB, on the CIFAR-10 dataset. Our results gave us comparable accuracy with respect to the other deep networks. Further details can be found in the report here.

Regression Analysis to Determine Effort Required for Software Maintenance OCTOBER 2014
Carnegie Mellon University, Pittsburgh

I used Hypothesis Testing and Regression Analysis to make decisions on Software development practices and how various independent variables ( like "software complexity","Project Team Experience", "Development Tools", etc) affects a software maintenance time.

Web Application for Virtual City GuideAPRIL 2014
Sardar Vallabhbhai National Institute of Technology, Surat

I worked towards integrating various web services such as web crawling and social media APIs, such as Twitter API and Facebook API. My team and I used these technologies along with databases to create personalized meta search engine with goal of listing all the basic and specialized directory results. The final product was a web portal which would allow the user to access all necessary information she/he would need while moving and settling in a new city.

Genome Sequencing using Hidden Markov ModelDECEMBER 2013
Sardar Vallabhbhai National Institute of Technology, Surat

I worked under the guidance of Dr. M. A. Zaveri and carried out a literature survey on Hiddern Markov Model. As part of my undergraduate requirements, I presented a seminar of HMM and its application in Genome Sequencing. The presentation can be found here.


Admissions Ambassador, Volunteer PositionMARCH 2015 - present
Carnegie Mellon University, Heinz College

I am one of the Admissions Ambassador for Heinz College - CMU. I interact with potential masters candidates as well as accepted students to help make them their decision regarding graduate schools. I aid them by providing all necessary information regarding the school and my experiences so that they can make an informed decision. Apart from this, I was part of the Career panel setup for visiting students.

Career Services Host, Volunteer PositionJANUARY 2015 - present
Carnegie Mellon University, Heinz College

I am a Career Services Host for Heinz College. My primary responsibilities are

  • Coordinating with company officials on the day of their Information Session.
  • Help setup the auditorium, for the Tech Talks.
  • Keep track of the students attending the event and be available for any queries.

Training and Placement Secretary, Students' SenateMAY 2013 - APRIL 2014
Sardar Vallabhbhai National Institute of Technology, Surat

I was selected as the 1st ever Training and Placement Secretary of the Institute. Responsibilities included but were not limited to

  • Coordinating with company officials.
  • Managing the budget allocated by the Institute for the Placement activities.
  • Organising mock interviews and tests for final and pre-final year students.
  • Point of contact for Campus activities organized by different companies like Mahindra AutoQ, Cairn India ACE 2 etc.

Joint Magazine Secretary, Students' SenateMAY 2012 - APRIL 2013
Sardar Vallabhbhai National Institute of Technology, Surat

My responsibilities, as the Joint Magazine Secretary,included managing the annual Institute Magazine, AS WE ARE. My duties consisted of

  • Organizing literary competitions
  • Editing articles and dealing with the publishers
  • Managing budget and presenting it to the students' senate
  • Distribution of the magazines to over 5000 faculties and students

Design Team Coordinator, MINDBENDJULY 2012 - OCTOBER 2012
Sardar Vallabhbhai National Institute of Technology, Surat

As the coordinator for Designing Team for the Institute Technical Festival, my responsibilities included supervising junior designer as well as creating event posters to promote the festival across to country, in other Institutes.

Magazine Secretary, Hostel Management CommitteeJULY 2010 - MAY 2012
Sardar Vallabhbhai National Institute of Technology, Surat

As the magazine secretary, I was responsible for all the hostel library and the news paper circulation of the hostel. Apart from this I was also responsible mess menu and services. I also helped the committee organize several hostel events.

Junior Editor, RENESA, Institute NewsletterNOVEMBER 2010 - APRIL 2011
Sardar Vallabhbhai National Institute of Technology, Surat

I was one of the editors for Institute newsletter and my responsibilities included taking interviews and covering various institute events.


Bronze Medal for Academic ExcellenceMAY 2014
Sardar Vallabhbhai National Institute of Technology, Surat

Best Paper for Role of Women in TechnologyNOVEMBER 2011
IBM Technical Web Contest

The IBM Technical Web Contest 2011, was conducted to celebrate the completion of 10 years by IBM Software Labs and to emphasize the importance of diversity in IBM.
My team was awarded Special Recognition for Best Paper in “Role of Women in Technology”, on an idea to improve the emergency calling system with aim of improving security for women.
We got an opportunity to present our idea on wider forum through cutting edge technology of Webinar using WebEx. Further details can be found in the IBM India University Relations newsletter

Poster PresentationSEPTEMBER 2011
I-CARE 2011 & IBM Centennial Colloquia

As part of the celebration for their Centennial year, IBM Collaborative Academia Research Exchange (ICARE) had organized various university level competitions.
My team was the youngest amongst the top 48 teams which were selected to present a poster at the IBM Centennial Colloquia held at The Grand, New Delhi, on the track of Smart Cities, Empowering Women. The poster can be accessed from here and further details of the event can be found here

Special Recognition OCTOBER 2011
InterNIT Robotics Meet

My team was the only team to effectively complete the task of building two fully autonomous robots one mounted over the other, which can coordinate with each other using intelligent swarm robotics to extinguish fire.

First Prize for designing a Mechanical ClockFEBRUARY 2011
MINDBEND, Technical Festival of SVNIT

My team was awarded first prize for designing a mechanical clock using only potential energy, calibrated 1 second with accuracy of 97%.

First Prize for Photography Competition, Das VidaniyaFEBRUARY 2011
MINDBEND, Technical Festival of SVNIT

I won the first prize for capturing the spirit of the technical festival in 10 photographs.

  • Python
  • Java
  • R
  • C
  • SQL
  • HTML
  • Assembly Language
  • Shell scripting
  • Apache Hadoop
  • AWS
  • Microsoft SQL Server
  • Apache Mahout
  • Twitter API
  • Adobe Photoshop
  • Apache Cassandra
  • Microsoft Access





Created by BLACKTIE.CO