Manager Quality Assurance

Anurag Narra

Personal Details

Address: B-***, Euphoria, Suncity, Iblur village, Sarajapur

Telephone Number: +91 - 900*******, +91 - 970-***-****

E-mail Address: *****.******@*****.***

rd

Date of Birth: 3 April 1987

Marital Status: Single

Nationality: Indian

Education

June 2004 – April 2008

Coimbatore Institute of Technology, Coimbatore

Governing Board: Anna University

Degree: B.E. Computer Science & Engineering

Cumulative Grade Point: 8.4 / 10

Student Internship

Defense Research & Development Organization, Hyderabad

Guide: Dr.K.R.Ravindranthan, Joint Director for Regional Center for Military Airworthiness

Problem Statement:

Research into various test suite reduction techniques used for software testing. Design and build a

prototype which would optimize the cost incurred while testing and also ensure the solution is scalable.

Background:

For testing to be time effective, redundant tests should be effectively removed from the test suite without

compromising much of the original suite's coverage and fault detection properties. Advantages of test

suite reduction include reducing the cost of executing, validating an d managing the test suite once the

product evolves.

Results

Designed an algorithm which could generate an optimized set of test cases provided a CFG. Optimizations

were done by graph reduction for which we achieved an approximate 12% reduction in test suite

execution time for a sample set of 400 CFG's. The solution was made scalable by storing persistent data

into a database. If changes were made to a small portion of code, this data could be retrieved and the

tests which covered only the recent changes could be executed. This further increases efficiency over a

period of time in a software development cycle. Based on this a technical report was submitted at DRDO.

Work Experience

June 2008 – February 2011

Microsoft India R&D, Hyderabad

SDET – Online Search & Ad Center

Responsibilities:

Majority of the responsibilities were setting up standards for software performance analysis, building

profiling tools for our internal teams and designing automation test frameworks for quality assurance.

Responsibilities extended to product development & design including ownership of performance & End 2

End testing of the entire Ad Center Editorial pipeline.

I have experience in developing an entire eco-system of test tools and documentation surrounding any

given web & desktop based applications.

Product lines:

Contributed to the development process of the tools List Manager, Editorial Verification Desktop,

Contextual – Ad Human Relevance System (Cx-AdHRS), Ad block detection, Ad Predict Online/Offline,

Contextual Revenue & Relevance, Bing bar and Desk bar.

Editorial Verification Desktop, List manager and Cx-AdHRS are part of Microsoft’s internal crowdsourcing

initiatives.

Ad block detection is a standalone tool which parses web pages and retrieves ads which are later used for

computing Bing’s ads relevance metrics. Bing bar is a browser based tool which provides news, sports,

weather etc. feeds and updates to end users.

Contextual-Revenue & Relevance provides the backend algorithms for calculating the relevancy of Ads

shown on publisher web pages. Our team has also solved the famous P-Click problem.

September 2011 – June 2012

Department of Bio-Medical Sciences, University of California – San Diego

Research Assistant

Guide: Wendy Weber Chapman

Problem Statement:

Medical reports have to be annotated for purpose of clinical decision support systems which require

annotations for training, validation and improvising classifier accuracy.

Responsibilities:

My responsibilities include designing experiments which would provide sufficient statistical significance

for viability of crowd sourcing when applied to solving the problems of Bio -Medical domain. During my

tenure I spent most of the time familiarizing and understanding crowd-sourcing literature. I have read

close to 160 journals on crowdsourcing which have given me a world of insight on the subject.

The idea was to architect a system which would crowd-source medical reports for annotations and the

challenges to tackle include task decomposition and workflow, validating accuracy of annotations, job

scheduling, mapping annotator’s accuracy, document and task clustering, decreasing cognitive load on

annotators and increasing motivation.

In the process I have been exposed to various statistical models used in economics, operations research

and other industries to optimally solve the above challenges.

Results:

Conceptualized the Bio-Medical sciences community platform which allows disease victims and their

families to annotate medical records.

Currently authoring a review paper/book on crowdsourcing which shall address crowdsourcing as a

science, the current trends and various other aspects of how crowdsourcing is a means to build collective

intelligence contrary to the popular belief.

Computer Proficiency

OS Platforms: Windows, Linux

Languages: C, C++, Java, Python, C#, MATLAB, SQL, R

Awards:

October 2010

Gold Star Award – OSD, Microsoft R&D

The Gold Star is awarded to a single individual every quarter in the entire Online Search Division -

Microsoft. This award was given in recognition to my contributions to Editorial Verification Desktop (EVD)

for the BOSTON release and specifically because I performance tuned EVD in the production

environments.

September 2009 & September 2010

Exceeding Potential rating awarded at Microsoft R&D.

Contact this candidate