Vijay Pappu

Ph.D. Candidate in Operations Research with expertise in Machine Learning at University of Florida

ab9i78@r.postjobfree.com

Summary

To obtain a full-time position as data scientist to work on real-world machine learning applications.

Specialties:

1. Three years of experience in building supervised learning models for classification and feature selection on

high dimensional datasets.

2. Three years of experience in statistical data analysis.

3. 5 years of programming experience in Matlab.

4. One year of programming experience in Java.

5. Two years of experience in high performance computing.

6. Strong analytical and presentation skills.

7. Strong communication and leadership skills.

Experience

Graduate Research Assistant at University of Florida

August 2010 - Present (3 years)

1. Worked on several projects related to developing supervised machine learning techniques for classification

problems arising in the field of biomedicine.

2. Gained expertise in understanding and applying several state-of-the-art methods like Support Vector

Machines, Neural Networks, k-Nearest Neighbor Classification, Random Forests, AdaBoost, Linear

Regression, Logistic Regression and Linear Discriminant Analysis.

3. Possess working knowledge of clustering techniques like Spectral Clustering, k-means clustering and

biclustering; and dimensionality reduction techniques like Principal Component Analysis, Local Linear

Embedding, Isomaps and Laplacian Eigenmaps.

2 recommendations available upon request

Graduate Teaching Assistant, Data Analysis and Data Mining in Systems Engineering at University of

Florida

January 2013 - May 2013 (5 months)

1. Conducted office hours for students, prepared and graded assignments and exams.

2. Assisted students in their course projects mainly selected from Kaggle competitions.

3. Taught invited lectures on Principal Component Analysis and Feature Selection techniques.

Data Science Research Intern at AddThis

June 2012 - August 2012 (3 months)

1. Worked on building a content based recommendation system based on user's web browsing history.

Page1

2. Developed a new webpage ranking algorithm that accounts for user's browsing history and the current

trends on internet.

3. Involved in all stages of product development including design, implementation, testing and automation.

4. Built an API for internal company feedback and further incorporated the changes to improve the product.

President, Gator Cricket Club at University of Florida

August 2011 - May 2012 (10 months)

1. Coordinated among officers/members of GCC to organize several tournaments, community service

activities and fund raising events.

2. Responsible for fund raising $1700 to facilitate activities of GCC.

3. Promoted GCC in several media outlets including Alligator, Gainesville Sun, PBS Gainesville and

American College Cricket.

4. Instrumental in achieving the Orange level (highest classification level) for GCC in 2012.

1 recommendation available upon request

Captain, University of Florida Cricket team at University of Florida

August 2011 - May 2012 (10 months)

1. Won the SE Regional Championship conducted by American College Cricket.

2. Responsible for improving the national ranking of UF cricket team from 13th to 2nd.

3. Won the Fall and Spring Cricket tournaments conducted by Gator Cricket Club at UF.

Application Support Engineer at The MathWorks

October 2008 - August 2010 (1 year 11 months)

1. Provided technical support for Matlab and Math products like Statistics, Optimization, Curve fitting etc.

2. Worked with the Optimization team to include new unit tests for the Optimization toolbox functions.

3. Promoted to a Specialist in Training (SIT) in Matlab/Math team within 1 year of joining.

1 recommendation available upon request

Instructor, Matrix and Numerical Methods in Systems Engineering at University of Florida

May 2010 - July 2010 (3 months)

1. Taught the course to seniors in Industrial and Systems Engineering and conducted weekly Matlab sessions.

2. Overall Instructor rating: 4.5/5.

Graduate Research Assistant at Rutgers University

August 2006 - August 2008 (2 years 1 month)

Dissertation: "Three Dimensional Computational Modeling and Simulation of Cell Rolling and Deformation

on an Adhesive Surface in Shear Flow"

1. Developed a 3D model based on Monte Carlo methods which simulates the stop-and-go motion of

leukocytes observed in-vivo.

2. Studied the effect of cell deformability, shear rate and cell concentration on rolling characteristics of

leukocytes during their adhesion process.

3. Performed all the simulations on National Center for Supercomputing Applications (NCSA) at the

University of Illinois at Urbana Champaign.

Page2

Projects

Sparse proximal support vector machines for feature selection in high dimensional datasets

August 2010 to Present

Members:Vijay Pappu, Pando Georgiev, Panos Pardalos, Paul Thottakkara

1. Developed a new embedded feature selection method that extends proximal support vector machines to

perform feature selection.

2. Proposed an efficient algorithm based on alternate optimization techniques.

3. Achieved more than 98\% feature removal rate without compromising on generalization performance for

high dimensional datasets.

4. Showed stability in the feature selection process in comparison with other univariate filter techniques.

On extending Principal Component Analysis (PCA) to feature selection using l2,1-norm minimization

February 2013 to Present

Members:Vijay Pappu, Paul Thottakkara, Pando Georgiev, Panos Pardalos

1. Proposed a new iterative algorithm to the problem of joint feature selection and subspace learning, that

controls sparsity and avoids overfitting.

2. Introduced heuristics in the proposed algorithm that achieves 100X speedup in running time for high

dimensional datasets.

Constrained subspace classifiers for high dimensional datasets

February 2013 to Present

Members:Vijay Pappu, Orestis Panagopoulos, Pando Georgiev, Panos Pardalos

1. Developed a new binary classification model that finds one subspace for each class while accounting for

relative orientation among the subspaces.

2. Proposed an iterative minimization algorithm based on alternate optimization techniques.

3. Achieved an improvement of approx. 10% over the local subspace classifier on several publicly available

datasets.

Least squares approach for Proximal Support Vector Machines

October 2012 to March 2013

Members:Vijay Pappu, Paul Thottakkara, Pando Georgiev, Panos Pardalos

1. Proposed a new least squares formulation for Proximal Support Vector Machines that would facilitate

feature selection.

2. Showed the equivalence among different formulations by performing numerical experiments on several

publicly available datasets.

Fisher-based feature selection combined with support vector machines for characterization of breast cell

lines using Raman Spectroscopy

October 2011 to October 2012

Members:Vijay Pappu, Michael Fenn, Ph.D., Pando Georgiev, Panos Pardalos

Page3

1. Developed an integrated framework that combines hierarchical clustering with supervised learning

techniques to perform feature selection and classification on Raman spectral data of five breast cell lines.

2. Achieved more than 95% classification accuracies utilizing only ten biologically relevant features out of

1200 features.

Alter Ego – A human centered model to provide counterfactuals

August 2011 to December 2011

Members:Vijay Pappu, Amey Deshpande

1. The project aims to provide counterfactuals (in terms of alternate professions) to individuals through a

human centered model.

2. An online survey was created and distributed among friends, colleagues and seniors belonging to different

professions.

3. The collected data was analyzed using two novel techniques namely Frobenius-Norm based Clustering and

Weighed Feature Selection methods to understand the nature of the data.

4. The internal validation on the data shows that these techniques work well and predict the professions with

high accuracy.

A simultaneous nonlinear least squares method for accurate characterization of polyhydroxy fullerenes

December 2010 to December 2012

Members:Vijay Pappu, Angelina T. Georgieva, Pando Georgiev, Panos Pardalos

1. Developed a simultaneous nonlinear least squares (SNLS) model for accurate characterization of

polyhydroxy fullerenes.

2. Proposed an algorithm based on scatter search and trust-region method to solve the SNLS model.

3. Estimated the number of hydroxyl groups for polyhydroxy fullerenes to be between 16 and 18 allylic

hydroxyl groups.

Skills & Expertise

Matlab

C

Fortran

Mathematical Modeling

Operations Research

Machine Learning

Data Mining

Optimization

Statistical Modeling

Java

Recommender Systems

Statistics

Strategic Leadership

Data Analysis

Page4

Publications

High dimensional data classification

Clusters, Orders and Trees: Methods and Applications

Authors: Vijay Pappu, Panos Pardalos

Recently, high dimensional classification problems have been ubiquitous due to sig- nificant advances in

technology. High dimensionality poses significant statistical chal- lenges and renders many traditional

classification algorithms impractical to use. In this chapter, we present a comprehensive overview of different

classifiers that have been highly successful in handling high dimensional data classification problems. We

start with popular methods such as Support Vector Machines and variants of discrim- inant functions and

discuss in detail their applications and modifications to several problems in high dimensional settings. We

also examine regularization techniques and their integration to several existing algorithms. We then discuss

more recent methods namely the hybrid classifiers and the ensemble classifiers. Feature selection techniques,

as a part of hybrid classifiers, are introduced and their relative merits and drawbacks are examined. Lastly, we

describe AdaBoost and Random Forests in the ensemble classifiers and discuss their recent surge as useful

algorithms for solving high dimensional data problems.

Sparse Proximal Support Vector Machines for feature selection in high dimensional datasets

Computational Optimization and Applications

Authors: Vijay Pappu, paul thottakkara, Pando Georgiev, Panos Pardalos

A new embedded feature selection method for high dimensional datasets is introduced by incorporating

sparsity in Proximal Support Vector Machines (PSVMs). Our method called Sparse Proximal Support Vector

Ma- chines (sPSVMs) learns a sparse representation of PSVMs by first casting it as an equivalent least

squares problem and then introducing the l1-norm for sparsity. An efficient algorithm based on alternate

optimization techniques is proposed. Numerical experiments on several publicly available datasets show that

our proposed method gives competitive or better performance compared with other embedded feature

selection methods. Moreover, sPSVMs remove more than 98% features in many high dimensional datasets

without compro- mising on generalization performance. Stability in the feature selection process of sPSVMs

is also studied and compared with other univariate filter techniques. Additionally, sPSVMs can be viewed as

inducing class-specific local sparsity instead of global sparsity like other embedded methods and thus offer

the advantage of interpreting the selected features in the context of the classes.

Raman spectroscopy utilizing Fisher-based feature selection combined with Support Vector Machines for

the characterization of breast cell lines

Journal of Raman Spectroscopy May 22, 2013

Authors: Vijay Pappu, Michael Fenn, Ph.D., Pando Georgiev, Panos Pardalos

Raman spectroscopy has the potential to significantly aid in the research and diagnosis of cancer. The

information dense, complex spectra generate massive datasets in which subtle correlations may provide

critical clues for biological analysis and pathological classification. Therefore, implementing advanced data

mining techniques is imperative for complete, rapid and accurate spectral processing. Numerous recent

studies have employed various data methods to Raman spectra for classification and biochemical analysis.

Although, as Raman datasets from biological specimens are often characterized by high dimensionality and

Page5

low sample numbers, many of these classification models are subject to overfitting. Furthermore, attempts to

reduce dimensionality result in transformed feature spaces making the biological evaluation of significant and

discriminative spectral features problematic. We have developed a novel data mining framework optimized

for Raman datasets, called Fisher-based Feature Selection Support Vector Machines (FFS-SVM). This

framework provides simultaneous supervised classification and user-defined Fisher criterion-based feature

selection, reducing overfitting and directly yielding significant wavenumbers from the original feature space.

Herein, we investigate five cancerous and non-cancerous breast cell lines using Raman microspectroscopy

and our unique FFS-SVM framework. Our framework classification performance is then compared to several

other frequently employed classification methods on four classification tasks. The four tasks were constructed

by an unsupervised clustering method yielding the four different categories of cell line groupings (e.g. cancer

vs non-cancer) studied. FFS-SVM achieves both high classification accuracies and the extraction of

biologically significant features. The top ten most discriminative features are discussed in terms of cell-type

specific biological relevance.

Feature selection based on meta-heuristics for biomedicine

Optimization Methods and Software

Authors: Vijay Pappu, Michael Fenn, Ph.D., Panos Pardalos, ##Ling Wang, Haoqi Ni, Ruixin Yang

Feature selection can e#ciently improve the accuracy of classi#cation and reduce the measurement, storage

and computation demands, and thus it has been applied in biomedical research increasingly. Considering the

NP-hard characteristic of feature selection, meta-heuristics are introduced into feature selection in

biomedicine on account of their excellent global search ability. However, most of biomedical problems are

characterized by high dimensionality, which is a challenge for feature selection methods based on

meta-heuristics due to the curse of dimensionality. Thus, six meta-heuristics, i.e. a Genetic Algorithm,

Particle Swarm Optimization, Ant Colony Optimization, Harmony Search, Di#erential Evolution, and

Quantum-inspired Evolutionary Algorithm, which are widely studied in the meta-heuristic community, are

introduced into feature selection in this paper and the performance of the algorithms is analyzed and

compared with each other for solving feature selection in biomedicine e#ectively. To evaluate the ability of

the algorithms fairly and exactly, a set of feature selection benchmark problems are designed and yielded for

the performance tests. The experimental results show that all the meta-heuristics are powerful enough to

achieve the ideal results on low-dimensional feature selection problems while it is essential to choose a

proper algorithm for the high-dimensional ones.

Least squares approach for Proximal Support Vector Machines

Optimization Methods and Software

Authors: Vijay Pappu, Paul Thottakkara, Pando Georgiev, Panos Pardalos

We introduce a new least squares formulation for Proximal Support Vector Machines (PSVMs), which

generates two non parallel proximal planes such that each plane is closest to one class while being

farthest from the other class. This formulation provides an opportunity to introduce sparsity into

the optimization model and thereby assist in feature extraction and feature selection.

Polyhydroxy fullerenes

Journal of Nanoparticle Research June 2013

Page6

Authors: Vijay Pappu, Angelina T. Georgieva, Pando Georgiev, Panos Pardalos

Characterization of C60 polyhydroxyfullerenes (PHF) prepared in alkaline media, preparation facilitated by

phase-transfer catalyst, presents challenges in determining the chemical structure resulting from the

possibility of multiple isomers or analogs with greater or fewer hydroxyl groups from a single reaction

mixture. This paper presents the utilization of analytical methods employed in tandem, especially X-ray

photoelectron spectroscopy, nuclear magnetic resonance spectroscopy, Fourier transform infrared

spectroscopy for semi-quantitative analysis on the number of hydroxyl groups present in PHF. Capillary

Electrophoresis was used for purity estimation of the material. Multiple spectra and electropherograms were

analyzed using a new simultaneous curve fitting method. The most accurate estimate of hydroxyl groups for

C60 polyhydroxy fullerenes obtained is between 16 and 18 allylic hydroxyl groups by combining analytical

methods’ results with 5 % accuracy. High precision (reproducibility) of the experiments is observed. Purity of

98 % is estimated by capillary electrophoresis. The size of PHF nanoparticles or aggregates has been

determined by atomic force microscopy to be 7.4–14.2 nm. According to the elemental analysis the average

probable empirical formula for the most pure PHF at pH 7.1 is C60O17H12Na5(NaHCO3)3(H2O)13 and the

average formula weight is 1,605.9 g/mol. This is the first thorough characterization of PHF in terms of purity.

Data Mining for Cancer Biomarkers with Raman Spectroscopy

Data Mining for Biomarker Discovery January 2012

Authors: Vijay Pappu, Michael Fenn, Panos M. Pardalos

Raman spectroscopy has the potential to play an important role in the diagnosis and treatment of cancer as a

unique type of biomarker technology. Raman spectra can provide a collective picture of the overall

composition of biological samples as well as highly sensitive targeting of speci#c biomolecular moieties

depending upon the application. In the #eld of Oncology, Raman Spectroscopy can help in the identi#cation

of biomarkers for use in drug discovery, cancer-risk assessment, histopathology, and in vivo clinical

applications. Continued advancements to data analysis techniques could prove vital in the realization of such

biomedical applications. This chapter provides a brief overview of some of the more common data analysis

methods as well as outlines several of the technical challenges encountered in the implementation of these

methods. The development of standardized data techniques with incorporation into fully functional integrated

software platforms will also be necessary for clinical applications in the future.

Data mining and optimization applied to Raman spectroscopy for oncology applications

BIOMAT -- 11th International Symposium on Mathematical and Computational Biology November 10, 2011

Authors: Vijay Pappu, Michael Fenn, Ph.D., Petros Xanthopoulos, Panos Pardalos

Recent advances in Raman spectroscopy have generated a surge of interest in biomedical applications

particularly in the field of oncology. As cancer is predicted to become the number one cause of death by the

end of the decade, Raman spectroscopy has the potential to significantly aid in the research, diagnosis and

treatment of cancer. Biomedical applications of Raman spectroscopy currently under investigation range from

the research laboratory bench-top to the clinical setting at the patients bedside. Raman spectroscopic analysis

of biological specimens is advantageous as it provides a spectral fingerprint, rich in molecular compositional

information without disrupting the biological environment allowing in-situ biochemical observations to be

made. The information dense spectra generate vast sets of complex data in which subtle variations may

provide critical clues in data interpretation. Thus the investigation and implementation of advanced data

Page7

mining and optimization techniques is imperative for complete, rapid and accurate data extraction. Clinical

applications of Raman spectroscopy are on the horizon as optical technology progresses, but to fulfill this

realization of a new class of biomedical instrumentation, the development of an optimized, fully integrated

data processing methodology will be required. In this paper, we describe several methods for pre-processing

raw Raman Spectra, followed by data mining techniques used for classifying spectra of cancerous cells based

on cell type, environments and mechanisms of cell death.

Hydrodynamic interaction between erythrocytes and leukocytes affects the rheology of blood in

microvessel

Biorheology September 10, 2007

Authors: Vijay Pappu, Prosenjit Bagchi

Hydrodynamic interaction between erythrocytes (RBC) and leukocytes (WBC) in a microvessel of size 20–40

micron, typical of a postcapillary venule, is studied using a two-dimensional computational model. The model

is based on immersed boundary method, and it takes into consideration the particulate nature of blood by

explicitly modeling individual blood cell, and cell deformation. Due to their highly flexible nature, RBC drift

away from the wall and toward the center of a vessel creating a cell-free layer. It is shown here that the lateral

motion of RBC is strongly affected in presence of a WBC, and is dependent on whether the WBC is

non-adherent or firmly adhered. When the WBC is non-adherent, some RBC, depending on their initial radial

locations and vessel size, may be deflected closer toward the wall, resulting in a decrease in the cell-free

layer. The apparent viscosity of the whole blood containing both RBC and WBC is computed, and shown to

be much higher than that containing RBC only. The increased viscosity cannot be accounted for by the

contribution due to WBC only. This observation is in agreement with a previous in vivo measurement. Here

we show that the additional flow resistance is due to the decrease in the cell-free layer resulting from the

WBC-RBC interaction. It can be accounted for by a two-layer model of blood when the reduced values of the

cell-free layer thickness are used. When the WBC is firmly adhered, RBC easily move away from the wall,

and the cell-free layer is not significantly changed. In such cases, the major contribution to whole blood

viscosity comes from the WBC alone. The hydrodynamic interaction between WBC and RBC, though it

exists, does not contribute significantly when WBC are adhered.

A computational study of leukocyte adhesion and its effect on flow pattern in microvessels

Journal of Theoretical Biology May 29, 2008

Authors: Vijay Pappu, Sai Doddi, Prosenjit Bagchi

Three-dimensional computational modeling and simulation are presented on the adhesive rolling of

deformable leukocytes over a P-selectin coated surface in parabolic shear flow in microchannels. The

computational model is based on the immersed boundary method for cell deformation and Monte Carlo

simulation for receptor/ligand interaction. The simulations are continued for at least 1 s of leukocyte rolling

during which the instantaneous quantities such as cell deformation index, cell/substrate contact area, and fluid

drag remain statistically stationary. The characteristic ‘stop-and-go’ motion of rolling leukocytes, and the

‘tear-drop’ shape of adherent leukocytes as observed in experiments are reproduced by the simulations. We

first consider the role of cell deformation and cell concentration on rolling characteristics. We observe that

compliant cells roll slower and more stably than rigid cells. Our simulations agree with previous in vivo

observation that the hydrodynamic interactions between nearby leukocytes affect cell rolling, and that the

Page8

rolling velocity decreases inversely with the separation distance, irrespective of cell deformability. We also

find that cell deformation decreases, and the cells roll more stably with reduced velocity fluctuation, as the

cell concentration is increased. However, the effect of nearby cells on the rolling characteristics is found to be

more significant for rigid cells than compliant cells. We then address the effect of cell deformability and

rolling velocity on the flow resistance due to, and the fluid drag on, adherent leukocytes. While several earlier

computational works have addressed this problem, two key features of leukocyte adhesion, such as cell

deformation and rolling, were often neglected. Our results suggest that neglecting cell deformability and

rolling velocity may significantly overpredict the flow resistance and drag force.

3D computational modeling and simulation of leukocyte rolling adhesion and deformation

Computers in Biology and Medicine June 2008

Authors: Vijay Pappu, Prosenjit Bagchi

A 3D computational fluid dynamic (CFD) model is presented to simulate transient rolling adhesion and

deformation of leukocytes over a P-selectin coated surface in shear flow. The computational model is based

on immersed boundary method for cell deformation, and stochastic Monte Carlo simulation for

receptor/ligand interaction. The model is shown to predict the characteristic ‘stop-and-go’ motion of rolling

leukocytes. Here we examine the effect of cell deformation, shear rate, and microvilli distribution on the

rolling characteristics. Comparison with experimental measurements is presented throughout the article. We

observe that compliant cells roll more stably, and have longer pause times due to reduced bond force and

increased bond lifetime. Microvilli presentation is shown to affect rolling characteristics by altering the step

size, but not pause times. Our simulations predict a significant sideway motion of the cell arising purely due

to receptor/ligand interaction, and discrete nature of microvilli distribution. Adhesion is seen to occur via

multiple tethers, each of which forms multiple selectin bonds, but often one tether is sufficient to support

rolling. The adhesion force is concentrated in only 1–3 tethered microvilli in the rear-most part of a cell. We

also observe that the number of selectin bonds that hold the cell effectively against hydrodynamic shear is

significantly less than the total adhesion bonds formed between a cell and the substrate. The force loading on

individual microvillus and selectin bond is not continuous, rather occurs in steps. Further, we find that the

peak force on a tethered microvillus is much higher than that measured to cause tether extrusion.

Courses

Bachelor of Technology (B.Tech.), Mechanical

Engineering

Indian Institute of Technology, Madras

Operations Research - I

Operations Research - II

Doctor of Philosophy (Ph.D.), Operations Research

University of Florida

Machine Learning CAP6610

Advanced Machine Learning CAP6617

Page9

Analysis of Algorithms COT5405

Fundamentals of Mathematical Programming ESI6912

Numerical Optimization MAP6208

Global Optimization EIN6918

Linear Programming and Network Flow Optimization ESI6417

Stochastic Modeling and Analysis ESI6546

Applied Probability Methods in Engineering ESI6321

Financial Accounting ACG5005

Finance-I, Asset Value, Risk and Return FIN5437

Managerial Economics ECP5702

Professional Writing GEB5212

Organizational Behavior MAN5246

Managers and Legal Environment of Business BUL5811

Global Strategic Management MAN6636

Production and Operations Management MAN5502

Managerial Accounting ACG5075

Master of Science (M.S.), Mechanical & Aerospace

Engineering

Rutgers, The State University of New Jersey-New

Brunswick

Fluid Mechanics 1 530

Fluid Mechanics 2 630

Computational Fluid Mechanics 534

Mechanics of Continua 554

Conduction Heat Transfer 570

Convection Heat Transfer 578

Computational Solid Mechanics 660

Methods in Applied Mathematics 527

Education

University of Florida

Doctor of Philosophy (Ph.D.), Operations Research, 2010 - 2013

Grade: 3.7

Activities and Societies: INFORMS, SIAM, IEEE

6 recommendations available upon request

University of Florida

Master of Science (M.S.), Industrial and Systems Engineering, 2010 - 2011

Grade: 3.85

Page10

Rutgers, The State University of New Jersey-New Brunswick

Master of Science (M.S.), Mechanical & Aerospace Engineering, 2006 - 2008

Grade: 4.0

1 recommendation available upon request

Indian Institute of Technology, Madras

Bachelor of Technology (B.Tech.), Mechanical Engineering, 2002 - 2006

Grade: 3.8/4.0

St. Aloysius High School

1999 - 2001

Languages

English (Full professional proficiency)

Hindi (Native or bilingual proficiency)

Telugu (Native or bilingual proficiency)

Honors and Awards

1. Honorable Mention Award - Seth Bonder Scholarship for Applied Operations Research in Health Services,

INFORMS, 2012-2013.

2. Outstanding International Student Award - University of Florida, 2011-2012.

3. Scholar Athlete of the year - University of Florida, 2011-2012.

4. President - Gator Cricket Club, University of Florida, 2011-2012.

5. Captain - Cricket team, University of Florida, 2011-2012.

6. Graduate Assistantship - University of Florida, 2010-Present.

7. Graduate Assistantship - Rutgers, The State University of New Jersey, 2006-2008.

Interests

Big Data, Hadoop, MapReduce, Mahout, NoSQL, Cricket,

Page11

Vijay Pappu

Ph.D. Candidate in Operations Research with expertise in Machine Learning at University of Florida

ab9i78@r.postjobfree.com

11 people have recommended Vijay

"I have known Vijay Pappu now for over 2 years as colleague and friend at the Center for Applied

Optimization at the University Of Florida. Vijay and I have worked together on several projects related to

data mining of Raman spectral data for my research based on breast cancer cells. I have also helped work

with Vijay on several other biomedical associated projects, including analysis of massive datasets collected

from DNA microarrays, NMR and mass spectrometry data collected from biological specimens. Vijay and I

have published two journal articles, two book chapters, and have presented our work at several prestigious

conferences. Vijay has strong work ethic, ability to rapidly learn and apply new knowledge and techniques, as

well as being capable of taking on the role as a team leader or team member in a variety of highly diverse,

multidisciplinary research settings. He is comfortable and capable of working in areas different than that of

his formal training and excelling in these areas. His mastery of his skills and knowledge in machine learning,

feature selection, optimization and many other areas of data mining and analysis of massive biomedical

datasets have been an exceptionally valuable asset to my work and our group’s collaborative efforts. Vijay is

highly knowledgeable in both the in the theoretical aspects of data mining as well as the practical applications

of data mining and analysis for a wide-range of biomedical data investigations. Vijay has a skillful awareness

for the importance of understanding the origins and actual implications of the data under analysis. Not only

has Vijay developed methods which are highly effective for Raman Spectroscopic data analysis and other

biomedical datasets, but also for a variety of large or massive datasets including finance, search, and image

classification/analysis, among others. Vijay is not only an expert in machine learning, but is also a great

communicator, and, importantly, a great teacher. I personally, have learned a tremendous amount about data

mining and machine learning from him. Vijay is one of the finest individuals I have personally had the

pleasure to have worked with in my career. He has tremendous potential for achieving greatness in many

areas, with his remarkable inter-personal skills, team leadership skills, and his expertise in data mining and

machine learning. I believe he would make an outstanding addition to any company or group. If you have any

questions regarding Vijay, his background or research, you may contact me at any time via email or phone for

further reference information about Vijay Pappu. Michael B. Fenn, Ph.D. Assistant Professor Co-Director of

the Center for Medical Materials and Biophotonics Department of Biomedical Engineering Florida Institute

of Technology 150 W. University Blvd. Melbourne, Florida 32901 Phone: 352-***-**** Email:

ab9i78@r.postjobfree.com"

Michael Fenn, Ph.D., Ph.D. Research Assistant, J. Crayton Pruitt Family Department of Biomedical

Engineering at University of Florda, worked directly with Vijay at University of Florida

"Vijay is an expert in understanding & analyzing data and is well versed with different machine learning

Page12

algorithms and methods. He is also an experienced programmer with many years of experience in Java and

Matlab. Not only is he technically proficient and knowledgeable, he is also a good team player."

Paul Francis Thottakkara, Instructor & Research Assistant, University of Florida, worked directly with

Vijay at University of Florida

"During my tenure as the treasurer of the Gator cricket, Vijay had a dual role of President and Captain. Even

with his action packed work, he would find time to work on the financial aspects and guide his team to the

SEC Championship victory. He made sure that the Club won the "Most Improved Club of The Year" at UF.

His achievements in academics and the club were then noticed by the UF officials who awarded him "Scholar

Athlete of the Year". It was great working with Vijay. He is a good leader and better team man. Good luck for

your future endeavors."

Dilip Shenoy, Student, University of Florida, reported to Vijay at University of Florida Gator Cricket

Club

"Vijay was great to work with on the MATLAB-Math team, perennially obsessed with applying machine

learning methods to improvements at work. He wore his enthusiasm for statistics and optimization on his

sleeve, and it was more than a little infectious on the rest of us. I'd work with him again in a heartbeat."

Manu Raghavan, Application Support Engineer, The Mathworks, worked directly with Vijay at The

MathWorks

"Vijay was my Teaching assistant for the Data Mining course. He took most of the classes for us and taught

many interesting topics in Data mining. His command on the subject and his way of teaching enabled me to

understand the topics better. He is definitely one of the sought-after persons to ask for when it comes to

getting the solutions to the complex mathematical problems. I believe his hard work, subject knowledge can

surely be an asset to any organization he works for."

Krishna Chaitanya, Student, University of Florida, studied with Vijay at University of Florida

"Vijay is enthusiastic about machine learning and data mining. He has rigorous thinking, great sense of

responsibility, and excellent programming skills. As the leader of UF cricket team, his communication and

leadership skills are also unquestionable. I highly recommend him."

Kun Zhao, Ph.D candidate, University of Florida, studied with Vijay at University of Florida

"He is a gem of a person, both personally and professionally. He is an excellent combination of hard work

and brains. He has achieved highs in academics, sports and relationship management at the same time. Any

institute, company and research center would be benefited to have a result oriented and team person like him

in their team."

Vinay Mansinghka, Graduate Student, University of Florida, studied with Vijay at University of Florida

"I know Vijay as graduate researcher and as a member of cricket team here at University of Florida. Vijay has

Page13

inspired me with not only his strong leadership skills as a captain of the cricket team, but also with his quality

to efficiently manage time by participating in sports beside having very strong academic qualifications. He is

very organized and a highly motivated individual to work with."

Anuj Goyal, Student, University of Florida, studied with Vijay at University of Florida

"I first met Vijay in the data mining course where he was the teaching assistant. He came across as a very

proactive and enthusiastic person. His communication skills and knowledge about data mining subject is

commendable. Even me who didn't have a very strong mathematical background found his way of teaching

very interesting and easy to understand. The assignments and exams designed by him was very practical and

extremely useful in learning such a complex subject. His enthusiasm to teach and make the students

understand shows his passion for the subject I'm confident that Vijay with a good blend of intelligence and

managerial qualities would be an asset to any organization."

Abhijit Rajan, Student, University of Florida, studied with Vijay at University of Florida

"Vijay is a very proactive and an enthusiastic individual with a voracious appetite for excellence. Having

known him for over 2 years now, both on and off the cricket field, I can affirm his excellent leadership and

communication skills. His achievements as both the President and the Captain of the Gator Cricket Team

stand as a testimony to his managerial and leadership skills. He is a meticulous planner, a goal oriented

individual and a clear thinker. He has a friendly, balanced personality with strong morals and is an amazing

guy to hang out with."

Aditya kumar Kasinadhuni, Research Assistant, Mareci Research Group, studied with Vijay at

University of Florida

"Vijay is a bright and energetic young scholar with leadership qualities. I had a chance to work closely with

Vijay in the Fluid Mechanics Lab at Rutgers University. During this time, I got a chance to see his

exceptional work ethic and a strong go-getter attitude. Besides being an intellectual, he has a charismatic

personality that would make a great team player. He has a diverse talent base, and is terrific in sports too. I

am confident that he will be very successful in his future endeavors and can assure that he will be a great

asset to any organisation."

Sai Doddi, Research Assistant, Rutgers University, studied with Vijay at Rutgers, The State University of

New Jersey-New Brunswick

Contact Vijay on LinkedIn

Page14

Contact this candidate