anatoli  [at]  aquery.com         Playa Del Rey,  CA  

Scientist, software architect, software engineer and mathematician with major experience in current web based architectures and algorithm development coupled with a very extensive research background. Experience ranges from on-line advertising, computer generated content, personalization, product recommendation, segmentation, data mining, high-volume text understanding to SEO and search engine algorithms. 

 

2005–2009      Oversee.net                 Los Angeles, CA

Chief Scientist  

With a team of researches developing innovative algorithms in different areas: 

Optimization of online document content and design; Algorithms of segmentation for testing;  Optimization of testing;  Behavioral targeting;  Computer generated content;  Algorithms for measuring correlations between linguistic objects;  Data mining;  Domain names understanding and categorization; User’s understanding and personalization;   Evaluation of semantic and syntactic similarity; Priming;  Query refinement;  Documents understanding and summarization.

2005–2005       Vendare Media          Sherman Oaks, CA 

Developed methodologies, algorithms, and prototypes for personalized and targeted advertisement and opt-in email personalization.  Application of Information Retrieval concepts to matching users and Websites.

2003–2004                  Amazon.com     Seattle, WA  

Personalization Scientist

Responsible for developing methodologies, algorithms, and  prototypes, aimed at increasing sales at Amazon by better targeting product recommendations to customers in the form of campaigns, direct email and online interactions.

The position draws upon my background in Data Mining, Information Retrieval, Pattern Matching and Search Technologies to build and interpret the databases needed to better understand: 

 ·   customers and their buying patterns,

 ·   characteristics of products

 ·   mapping between customers and products

Customer profiles are built by combining demographics, and behavioral patterns such as site navigation and buying history with detailed information about the products bought, to create a fairly robust profile given limited knowledge of the customer.

Product profiles are built starting with information extracted from vendor product descriptions and supplemented by intelligent agents, which extract relevant information from the Web to automatically create a more robust product profile. Some of the major technical challenges include, extracting information from unstructured data on the web, and handling of sparse matrices. The conceptual  challenges include in-depth understanding of the intricate mechanisms of collaborative filtering, association rules, and behavioral targeting. 

1999–2003                  aQuery, Inc          Somerville, MA 

Co-founder, President  / Research Scientist 

      AQuery has developed cutting edge technologies to locate unstructured data on the Web, understand the content, extract relevant information, and organize the results into structured reports. The product relies on a deeper understanding of documents and queries to improve the relevancy of search results. The recall and precision achieved significantly exceeded that of search engines (such as Fast, Alta Vista, Lycos), which primarily rely on keyword methods. The performance rivals that of search engines using popularity analysis -- such as Google, with the difference that the techniques I co-developed apply to situations with insufficient popularity information.  Our approach also allowed us to perform searches at a finer level than other search engines -- enabling sub-paragraph search of coherent sets of several adjacent sentences. aQuery has developed a condensed categorized presentation of search results, combining text, images, tables, and advertisements in one document.  Preliminary user testing has shown that this presentation format outperforms the standard list of short descriptions of Web documents returned by a conventional search engine. The technology can also intelligently classify documents by their intent (i.e. Sell a product, Give advice, Provide a review, FAQ, etc.), and uses distributed agent technology for performing topic-directed crawl of the World Wide Web.  The architecture and algorithms are designed to store, categorize, navigate, retrieve and present heterogeneous data and information (spanning both text and images).

My major management activities have include:

 ·   Complete management of the company.

 ·   Recruited, directed, and motivated a technical team of 5-12.

 ·   Wrote business plan that raised $1.6 million in financing from Accel Partners; and brought Mitch Kapor to company as Board Chairman.

 · Filed 2 patents.

Technical activities involved the development of architecture for all projects and included:

·   Oversaw development of distributed agent technology for performing topic-directed crawl of the World Wide Web, allowing rapid collection and refresh of information on the Web about a specific topic.

·   Designed document level and chunk level Search Engines.

·   Designed Search Engine which finds query related Tables  and/or Images.

·   Co-designed and oversaw implementation of algorithms and architecture to store, categorize, navigate, retrieve and present heterogeneous knowledge (spanning both text and images).

·   Designed novel knowledge mining and knowledge representation components.

·   Designed anti-spam technology.

·   Oversaw development of agent technology for intelligently classifying documents by their intent (Sell a product, Give advice, Provide a review, FAQ, etc. “detectors”).

·   Co-designed Precisely Directed Advertisement capability – a technology, which finds correspondence between advertisements and WEB documents.

Technology was successfully commercialized in applications for categorization and evaluation of email messages, and was successfully adapted for a customer developing a wide array of financial information.

1995–1997      Virtual Information Systems   Saint Paul, MN 

Systems Analyst

Team member consulting for major corporations including Honeywell and Eaton.

Oracle applications, interfacing with third party Oracle packages and legacy systems.  

 

 1991–1994      Digital Biometrics, Inc.        Minnetonka, MN

Software engineer

Developed applications to solve a wide variety of fingerprinting recognition and classification problems.  Major activities included: 

 ·   Designed and developed a variety of data compression algorithms. 

 ·   Formulated and implemented a variety of novel approaches to image processing algorithms.

 ·   Delivered a unique, highly effective solution of the fingerprint classification problems.

 ·   New algorithms for machine vision using original algorithms of the generation of templates.

 ·   Noise reduction algorithms.

 ·   Developed new approaches to generation fuzzy logic interfaces, modeling human decisions in classification of fingerprints.

 

1990–1991                  Ultimap, Inc.   Bloomington, MN

Software engineer

Designed and developed software for simple geometrical representation of complex polygons in Geographical Systems. Development in C/C++.

1990–1990       University of Minnesota    Minneapolis, MN

Research fellow

Designed and developed algorithms for signal detection in astronomical data, finding regular patterns in data containing polynomial noise.    Development in C.