Ankit Rakha

17 E. Sir Francis Drake Blvd. —— Suite 110 —— Larkspur, CA 94939

Phone: 410-402-0406 | GitHub page: https://github.com/ankit-rakha

E-Mail: ankit21@cs.jhu.edu

Education

o MSE in Computer Science, Johns Hopkins University, USA

Advisor - Prof. Mitra Basu & Edward J. Schaefer Professor S. Rao Kosaraju

o B.Tech in Electronics and Communication, MMET Deemed University, India

Publications & Patents

o Journal Articles:

Human Mutation 2011 - "Rapid and efficient human mutation detection using a bench-top next-generation

DNA sequencer" - Qian Jiang, Tychele Turner, Maria Sosa, Ankit Rakha, Stacey Arnold and Aravinda

Chakravarti

Circulation Research 2012 – “Effects of Rare and Common Blood Pressure Gene Variants on Essential

Hypertension: Results from the FBPP, CLUE and ARIC Studies” - Khanh-Dung H. Nguyen, Vasyl Pihur, Santhi K.

Ganesh, Ankit Rakha, Richard Cooper, Steven C. Hunt, Barry I. Freedman, Joe Coresh, Wen H. L. Kao, Alanna C.

Morrison, Eric Boerwinkle, Georg B. Ehret, Aravinda Chakravarti

o Patents:

A Method of detecting retained foreign bodies (RFBs) - Joint Collaboration with Johns Hopkins Bayview Medical

Center, Reference No. C10417 - Ralph Etienne-Cummings, Ankit Rakha, Bolanle Asiyanbola

o Poster Presentation:

RECOMB 2008 - “HLA class I peptides: Exploiting positional information for identification and classification” -

MitraBasu, Ankit Rakha

ISMB 2008 - “Identification of key amino acid properties for HLA Class I peptides: A Computational Approach” -

Ankit Rakha, Mitra Basu, Rao Kosaraju

Experience

o Big Data and Analytics Practice Lead, Lilien Systems

March 2014 – Present

1. Develop and implement Big Data and Analytics solutions which include designing, deploying, testing

and provide post-installation support.

Jan 2007 – April 2008

Master Thesis - “A Quantitative Study of Ensemble Characteristics of peptides that bind to MHC class I

molecule”

Sept 2001 – May 2005

Page 1 of 8

2

2. Work with the Lilien Infrastructure and Network Solution Architects to design and implement

comprehensive “soup to nuts” Analytic solutions, integrating into the customer’s infrastructure across the

data center.

3. Identify and scope the services necessary to implement solutions including the preparation of pricing

estimates, statements of work and RFP responses.

4. Set up vertica database for development environment in AWS and for production environment in data

center.

5. The vertica set up including schema creation, roles, users and permissions and offloading of data from

Oracle warehouse was automated using Qubell, integrated with F5 for load balancing, Tableau for

visualization and Talend for data integration.

6. Set up dev and production hadoop cluster using cloudera distribution.

7. Upgraded vertica production database from 6.x to 7.x version.

8. Designed and implemented the regular data roll out from vertica to hadoop cluster. We keep only 3

months of data in vertica and rest goes to Hadoop. The data in hadoop cluster is accessed as external

hive tables and on vertica side as external vertica tables.

9. Integrated R-Vertica, R-Hadoop: R-Impala & R-Spark, Vertica-AWS in order to perform metric and

predictive analytics on game data.

10. Architected and proposed a solution for activator discovery, profile, message and lead vendor

screening using state of the art machine learning algorithms for distributed computing frameworks. The

ecosystem consist of big data technologies like vertica, hadoop, distributed R and Spark.

11. Architected the hadoop cluster for their recent initiative to create a hybrid platform using Apache

Kafka, Vertica and Hadoop. Extensively involved in creating the specifications of the hadoop, vertica

and distributed messaging queue nodes while designing the infrastructure layer.

12. Architected and proposed a predictive real time analytics platform for churn prediction for first day

churners through first session experience, monetization alternatives for potentially churning player in the

first session, churn prediction through customer’s last experience of x sessions and many more use cases.

The ecosystem consists of a seamless integration of VoltDB, Vertica and Hadoop and the use of

machine learning algorithms like linear regression, k-means, SVMs, decision trees and Page Rank tailored

for the particular use cases.

o Software Engineer – Big Data Team, Serendipity (Parent company: GCE)

1) DevOps

Jan 2013 – March 2014

a) Maintained 25 cluster nodes / 1200 cores running Centos 6.4. Administration responsibilities included

assembling of the servers, installing OS using PXE, updating system software like upgrading gcc to

the latest versions using source code, installation of Vertica, Hadoop and MPI stack along with Solr.

b) Wrote custom build script with clean and fast options for application developers to build and deploy

the MPI project.

Page 2 of 8

3

c) Installed and used Chef for configuration management.

d) Also, provided support in migrating federal government datasets from Vertica database to HBase.

We wrote custom tools namely DataSync, DataImporter and DataProvider for this purpose.

2) Distributed computing using MPI

Worked on creating low latency distributed computing system for real time analytics. We used openMPI

1.7.2 and intel mpi 4.1.2 (for multi-threading support), C programming language and HBase as working

components for this project. Introduced multiple sub-projects like

a) Statistical algorithm service - Implemented plethora of statistical algorithms in distributed fashion

using MPI routines like MPI_Send, MPI_Recv, MPI_Reduce, MPI_Allreduce, MPI_Gather, MPI_Gatherv,

MPI_Bcast. Some of the implemented algorithms include: Numerical histogram, Categorical

histogram, Minmax, Kmeans, Regression, Pearson and Spearman correlation coefficient, Line

intersection algorithm, Percent distinct, Quantile, Anomaly detection, Median etc.

b) Data service – Worked with other team members to create a service that uses Apache Thrift API to

read data from HBase and create a memory store.

c) Instruction topology – Created multi-threaded topologies with execution plan to recommend

visualizations to users.

d) Graph service – Represented the hierarchical data in the form of graph nodes and edges to

implement zoom-in and zoom-out capabilities while analyzing the data.

3) Graph Database

Evaluated two graph databases – Neo4j and Titan for semantic analysis work. Migrated data from

Vertica database to the graph database so that information can be represented in the form of nodes

and edges. Used batch-import and parallel batch inserter tool for migration of data into Neo4j -

https://github.com/jexp/batch-import and faunus + HBase for Titan graph database.

o Software Engineer, Mckusick Nathans Institute of Genetic Medicine, Johns Hopkins University School of

Medicine

May 2008 – Dec 2012

A) Software development & Computational analysis

1) Responsible for implementing, documenting and maintaining software tools, scripts and codes.

2) Worked with biologists and statisticians in research environment to meet their dynamic programming

needs.

3) Development and applications of molecular genetic, genomic and computational methods for

identification of human disease genes. Dealt with data sets that are immense in size and developed

new and better algorithms to analyze them using parallel computing platform for development,

evaluation and application of statistical and computational methods on tens of thousands of genomic

arrays. The algorithms we used require large-scale memory access and lot of matrix manipulation. So

we utilized grids hosted in Johns Hopkins University and University of Maryland for data processing and

interactive simulations with very large, real-life data sets. We broke the computations (serial code) into

Page 3 of 8

4

independent tasks (parallel) as in parallelization of FOR loops or Monte Carlo Simulations for performing

analysis very fast.

B) IT consultation

1) Analyzed IT requirements and provided strategic guidance to research groups within Johns Hopkins

Hospital with regard to IT technology, cloud service infrastructures and distributed computing. This

included but not limited to introduction to Apache hadoop ecosystem, scale out storage and highperformance

enterprise file management platforms like EMC Isilon, OneFS, GPFS and GlusterFS.

2) Organized seminars and lead meetings with IT teams from Red Hat, Cloudera, Infochimps, IBM and

EMC.

3) Prepared documentation and presented progress reports to the principal investigators regarding new

distributed computing infrastructure and return on their investments.

C) Database development

1) Reviewed process design, stored procedures to correct errors and make necessary modifications in SQL

codes in internal database system.

2) Worked with lab members to write new database procedures, functions and triggers.

D) System administration

1) Interacted and closely worked with System administrator, LAN administrator & Programmer Analysts to

develop specifications for solutions that were ultimately implemented by the developer.

2) Maintained lab’s raw data sets repository, local web servers of publicly available and licensed

databases, remote access & internal server accounts for current lab members, archiving of old lab

members server data.

E) Web programming

1) Responsibilities included both front-end and back-end programming using different server-side

frameworks to build database driven websites, cross-browser compatibility testing, and search-engine

optimization tasks for deployed websites.

o Research Assistant, Computer Science Department, JHU

June 2007 - Aug 2007

Worked on Machine Learning techniques to characterize HLA peptide-binding predictions associated with

Arthritis disease. Closely worked with collaborators and research groups to develop streamlined processes,

workflows, and database applications for bioinformatics applications. Prepared and maintained

documentation to support use of the workflow and database applications developed. Also, created and

presented tutorials on bioinformatics, genomics and proteomic software and databases.

Skill Sets

o Programming Languages – C/C++ (excellent knowledge, several team projects of more than 25K lines of code)

Page 4 of 8

5

o Scripting languages – Unix shell scripting (well mastered), perl, sed & awk

o Statistical languages - MATLAB, R (heavily for plotting)

o Data Interchange formats: JSON, Google Protocol Buffers

o Data Integration tools: Talend, Syncsort DMX-h

o Visualization tools: Tableau, Yellowfin

o CMS – Drupal (have used version 6 & 7), Wordpress

o Database systems – HBase, Vertica, VoltDB, Riak, PostgreSQL, MySQL, Neo4j, Titan

o Version control system – git, SVN

o Editors/IDE – Xcode, Intellij, Netbeans, AptanaStudio3, TextMate, Sublime, Notepad++

o OS - Mac OS X, Red Hat Enterprise Linux (RHEL), CentOS, Linux Fedora, UBUNTU, Windows XP, 7 & 8

o Cloud computing

Parallel & distributed computing using SGE and TORQUE, Amazon EC2 for cluster based applications, used

Apache Hadoop, Hadoop Distributed File System & MapReduce for large datasets and number crunching

problems. Highly familiar with CLUSTERS hosted at following sites:

1) Sun Grid Engine hosted by IGM, Johns Hopkins University – School of Medicine

(https://paradigm.jhmi.edu/services/hpc)

2) Sun Grid Engine hosted by Biostat department, Johns Hopkins School of Public Health

(http://www.hpscc.jhsph.edu/)

3) Terascale Open-Source Resource and QUEue Manager hosted by Johns Hopkins Physics and

Astronomy Department (https://hhpc.idies.jhu.edu/wiki/index.php/Main_Page)

4) Data Instensive Academic Grid hosted by IGS, University of Maryland – School of Medicine

(http://diagcomputing.org/)

5) Amazon Web Services - http://aws.amazon.com

o Bioinformatics Softwares & Databases – PLINK, Galaxy, Haploview, Beagle, SAMtools, VCFtools, tabix, snpEff,

PolyPhen, SIFT, ANNOVAR, bigwig, EVS, GATK

Projects

DATABASE & WEB-APPLICATION DEVELOPMENT

o HDRC – A collaborative research website (https://hdrcstudy.org/)

Page 5 of 8

6

Building an early-stage collaborative research website using Drupal 7. It required hands-on development in

PHP, MySQL, HTML, JavaScript, and CSS. Performed browser testing, security audits, and stress tests. Maintaining

web analytics, conversion reporting, and log analysis tools, including Google Analytics and urchin tracking

code. Administered MySQL databases. Wrote command-line utilities and scheduled tasks using cron.

o LabWorks – Internal Database application (http://labase-web1.igm.jhmi.edu:8500/labworks/login/login.cfm)

Designed and implemented modules of a database web application using Adobe Coldfusion, Dreamweaver,

postgreSQL database & pgAdminIII with six major components and supports 11 postgreSQL databases. The

major components include enrolling new patients and their information in encrypted manner into the

database, organizing their DNA samples in hierarchical order, maintaining corresponding genotypes for all the

SNPs that were genotyped, location information of the samples in the cold storage, file uploader for keeping

the track of protocols that were used in genotyping phase and administration section for adding new groups

and members.

o ACLAB website – An iWeb project (https://aravindachakravartilab.org)

Designed and built a simple, clean and crisp customized website using Apple’s iweb software (3.0.3) and

JavaScripts.

o Request Tracking system – Ruby on rails project (http://alabwiki.jhmi.edu:3000/login)

Created components of a RTS using Ruby on rails framework to manage requests for administrative managers

efficiently. It provides a flexible role based access control to a group / lab.

o ARIC & FEHGAS Wiki system

(http://162.129.237.10/aric/index.php/Main_Page,http://162.129.237.10/jhu/index.php/Main_Page)

Maintaining online database wikis with MySQL database at the back end to share data and information

between collaborators of an ongoing research project.

SOFTWARE DEVELOPMENT

o HSCR Exome Sequencing Project – Data analysis and text mining using map-reduce paradigm

Currently analyzing exome data for Hirschsprung disease in 304 individuals (3.3 TB of binary data) using The

Genome Analysis Toolkit framework developed by Broad institute. We extensively take advantage of mapreduce

paradigm in Quality control of the data. The tools being used include various walkers like depth of

coverage analyzer, snp/indel caller and variant quality score recalibrator. Annotation part of the project

concentrates on using text mining techniques to extract the information about genes, proteins and their

functional relationships from html webpages, text files and backend databases.

o RSNG Sequencing Project – An automated pipeline for identification of potential heterozygotes

In this project, we implemented a computationally efficient linux computer program using customized unix shell

scripts, Matlab distributed computing server and various sub-components including parallel, bioinformatics and

statistics toolboxes that compares sequences from 11 genes across large number of amplicons obtained from

different individuals (no. of individuals = 560) to identify heterozygous sites for single nucleotide substitutions.

o GS Junior Sequencing Project – Rapid and efficient human mutation detection using a bench-top nextgeneration

DNA sequencer

Analyzed large sequencing data generated by bench-top next generation DNA sequencer. It involved

calculating the similarity between the consecutive runs for the same set of patients and determining the

Page 6 of 8

7

accuracy of the machine. Also generated a nice grid output for the users so that they can adjust their

amplicon pooling after each run.

o Image Processing Project – A Method of detecting retained foreign bodies (RFBs)

In this research project, we developed a model that detects if there is any surgical instrument left in the patient

body after an operation. This work involves SVM based method to classify X-ray images that have surgical

instruments left over.

o SLAM - Simultaneous Localization And Mapping (Cross Collaboration with NASA, UMD and UAF)

Worked in a collaborative enviornment for developing a computationally efficient sensing and modeling

method that uses images from one or more cameras on a robotic vehicle to build a world map, and to

accurately localize the robot within this map.

o Bioinformatics Project - Protein-Peptide Binding Prediction based on Learning Peptide Distance Functions

Designed & Implemented data mining & machine learning techniques for the classification of the peptides

binding to HLA protein using Multiclass Adaboost, Gaussian Mixture Models. The general approach can be

considered to be semi-supervised learning problem with partial information in the form of equivalence

constraints and also involves learning peptide-peptide distance functions.

o Machine Learning Project - Supervised & Unsupervised Learning for Cluster Analysis

Implemented algorithms that can discover high purity clusters in unsupervised, large, very noisy, highdimensional

datasets where most of the data points do not cluster well. Application domains include marketbasket

data.

o Robotics Project - RoboCup, Robot Soccer World Cup

Vision – Designed and Implemented Adaboost Algorithm with a variation of weighted nearest-neighbor

classifier for the weak learner part for classification and detection of soccer ball independent of its color, in the

presence of changing illumination, texture and partial occlusion.

Non-Profit & Volunteer Work

1) Lending loans to alleviate poverty through a non-profit organization named Kiva. Kiva works with

microfinance institutions on five continents to provide loans to people without access to traditional

banking systems. URL: http://www.kiva.org/lender/ankitrakha

2) Contributing member of a distributed computing project named SETI@HOME. SETI@home is a scientific

experiment that uses Internet-connected computers in the Search for Extraterrestrial Intelligence (SETI).

URL: http://setiathome.berkeley.edu/view_profile.php?userid=9795743

Keywords

Software & Web Applications Development, BIG DATA, Machine Learning, Statistical Analysis, Computational

Biology, Data Mining, Pattern Recognition, Artificial Intelligence, Bioinformatics, Multiple Classification &

Clustering Analysis

Page 7 of 8

8

Page 8 of 8