Sample Template Example of Beautiful Excellent Professional Curriculum Vitae / Resume / CV Format with Career Objective, Job Profile & Work Experience for Freshers & Experienced in Word / Doc / Pdf Free Download
Download Resume Format
ñMason Elberti
lSelected Accomplishments
Lead researcher for search algorithm
development on WestlawNext (www.WestlawNext.com), a state-of-the-art legal research application. Search algorithms blend traditional text
search, customer usage data, document citation patterns, and heterogeneous
metadata with machine learning; providing a quantum jump in retrieval
performance. WestlawNext was a major
multi-year initiative for Thomson Reuters and is positioned to supplant Westlaw
a $1.5 billion business worldwide. Was
one of ten member team selected from amongst Thomson Reuters 50K employees to
win 2010 President’s Award.
Responsible for text categorization
algorithms developed for Reuters Insider (About Reuters Insider), a video
content aggregation service providing financial professionals the tools to go
beyond the headlines to access in-depth analysis. The solution utilizes speech to text
technology on live market news feeds, allowing the application to then
topically categorize the news stories, extract and resolve named entities, and
locate industry keywords. Categorization
solution blended proprietary classification techniques and extracted metadata
with modified SVM to boost performance 20 points above baseline precision. Reuters Insider is the winner of a 2010 SIIA
CODiE award.
Lead classification research effort for
Contract Normalization product slated for production Q3 2011. Contract Normalization service automatically
identifies risky legal language in contracts drafted by legal professionals and
suggests similar clause passages.
Solution involves unsupervised free text clustering of 9 million
contracts on Hadoop cluster, as well as offline classification solution
implemented in Hadoop.
Lead team effort as principal developer
for interactive ‘Day In The Life’ visualization application that displays
customer interactions with Thomson Reuter’s platforms and content. Project involved a large data collection and
normalization component having to gather data from across heterogeneous
business units. Very successfully
learned and utilized Processing language to represent customer interactions
around the globe in both 2D and 3D renderings.
Both CEO and CTO have utilized application at conferences and a movie
version is slated for release to external website mid-2011. With absolutely no prior graphics or visual
design experience, project was completed in less than 3 months to meet iron
clad deadline.
Primary researcher and developer for
ResultsPlus (About ResultsPlus) an analytic recommendation engine for legal researchers, currently
responsible for $40 million in incremental revenue. Instrumental in both algorithm development
and scaling product from a handful of taxonomies to over three hundred and
fifty taxonomies, containing over 2 million articles for recommendation.
Pioneered usage of user click through
data in company (seven years ago). Daily
usage logs were loaded to databases.
With the help of brilliant team members was able to architect live
algorithm testing to enhance recommendation relevance and significantly
increase ResultsPlus revenue (~15%).
Integral member of small team that
successfully applied proprietary classification technology to enhance a broad
range of backend editorial processes at West.
lProfessional Experience
Lead Analytics Engineer – NGMOCO, San Francisco ? 2011 – Present
Responsible for design and implementation
of recommendation serving algorithms for games platform. Designed system to 'close the loop' to enable
collection of performance metrics for algorithms and allow for sophisticated
split A/B testing. Implemented churn
prediction models for mobile games.
Senior Research
Scientist – Thomson Corporate,
New York, NY ? 2009 – 2011
Hand-picked to open and lead new office
for corporate R&D in New York City, currently housing 9 of R&Ds 40+
researchers and developers.
Point person responsible for research
efforts involving financial unit (Thomson Reuters Markets). Project engagements range from short POCs and
feasibility studies to large multi-year product development involving dispersed
development teams.
Currently leading text categorization
effort for Contract Normalization.
Hands-on work includes experimentation and evaluation utilizing in-house
text categorization engine (jCaRE), implementing Java Map/Reduce methods for language
modeling corpus, and experiments utilizing SVM to merge heterogeneous feature
evidences for optimal classification performance.
Researcher/Developer for multiple text
categorization projects to include Reuters Insider (news transcripts),
SmartTerms (news articles), and RCS codes (financial documents) utilized
technologies include Java, DerbyDB, SVMLight.
Lead development of ‘Day in the Life’
visualization application that displays customer interactions on Thomson
Reuter’s products. Was primary developer,
and with no prior graphics experience learned Processing language and
implemented quality solution in 3 months.
Technologies utilized included Processing, Java, OpenGL, DerbyDB, GIMP.
Wrote algorithms in Perl to crawl and
scrape news sites to collect training text as input to semi-supervised
classification solutions.
ñSenior Research Scientist – Thomson Reuters Global Resources, Switzerland ? 2007 – 2009
Lead researcher for search algorithm
development on WestlawNext. Lead small
team of researchers designing and implementing next generation search
algorithms for flagship legal search engine.
Personally, implemented dozens of algorithms for document discovery and
ranking – tested, evaluated, and tuned hundreds of variants. Ranking algorithms involved heavy feature
engineering component across asymmetric metadata types. Final solution melded
evidence from traditional IR search features, citation patterns, user click
data, and document metadata to produce quantifiably superior search
performance. Primary technologies
utilized included Java, JDBC, Microsoft SQL server, and C.
Asset manager for search and clustering
IP, as such worked abroad at TRGR in Switzerland.
ñResearch Scientist –Thomson Legal & Regulatory R&D, Rochester, NY ? 2004 - 2007
Responsible for research and
development of core algorithms powering ResultsPlus legal recommender system on
WestLaw platform. Additionally
Implemented multithreaded (POSIX) splitter/collector service in C++ to distribute
and merge recommendation requests from client to a large array of backend
services.
Took personal responsibility for all
scaling issues on ResultsPlus system.
This included handling latency and throughput issues, service/server
configuration, hardware estimates, load balancing algorithms, throttling, and
troubleshooting the system. The system
has been scaled up to running on over 60 Windows servers, suggesting millions
of documents with sub second response times.
Pioneered idea of using user click
through data in conjunction with customer metadata to increase ResultsPlus
click through rates. Developed an array
of algorithms to optimally rank ResultsPlus recommendations and utilized live
A/B testing to validate best algorithms.
Optimization resulted in 15% increase in daily revenues.
ñSenior Software Engineer –Thomson Legal & Regulatory R&D, Rochester, NY ? 2001 - 2004
Evaluated, tuned, and implemented text
categorization solutions on over a dozen different legal taxonomies to improve
efficiency of editorial back office processes.
Most taxonomies had over 10k distinct categories and a couple exceeded
over 100k categories posing distinct software and classification performance
challenges. Experiments and evaluation
tools were written in C/C++ in a UNIX environment.
Part of team that developed proprietary
C++ text categorization engine (CaRE) on UNIX platform. Responsible for text processing algorithms as
well as development of inverted indices.
ñSoftware Engineer – Sungard Trading Systems, Pittsford, NY ? 1999 – 2001
C/C++ UNIX development of Universal
Market AccessTM product. A real time client/server application for
intelligent quote management and market making access. Included independent development and
integration of the NASDAQ SOES order system into existing software. Significant POSIX multithreaded development.
lMachine Learning/NLP Experience
Over ten years experience in algorithm
research to include machine learning, text categorization, IR, usage analytics,
data mining, business intelligence, pattern recognition, news flow algorithms,
data visualization. Extensive experience
with packages to include libSVM, SVMLight, Weka, Regression, Lucene, Lemur,
UIMA, CLUTO.
lTechnical Skills
Proficient Java, C, and C++ programmer,
especially adept at developing low-latency memory efficient algorithms. Significant experience with large-scale
multi-threaded client/server applications. Significant experience with SQL,
Oracle, BerkelyDB and a variety of other DB applications. Development work on both UNIX and Windows
platform. Additional experiences with
Perl, Python, and Processing languages.
lPapers & Patent Work
Khalid Al-Kofahi, Peter Jackson, Charles
Elberti & William Keenan (2007). A document recommendation
system blending retrieval and categorization technologies. Paper
presented at the AAAI Workshop on Recommender Systems in e-Commerce,
Vancouver, British Columbia, Canada.
Khalid Al-Kofahi, Michael Dahn, Patrick
Slaven, Thomas Zielund, Qiang Lu, Charles Elberti.
“Systems, methods, and software for identifying
relevant legal documents,” Thomson Corporate, U.S. Patent App. Number
11/538,749, October, 2006.
lEducation
State University of New York at Brockport,
NY, Mathematics ? 1995 - 1999
Winner or 1997 Adolph Lipinski award in
Mathematics.
lMilitary Experience
United States Active Army ? 1990 - 1994
Download Resume Format
0 comments:
Post a Comment