Bob Dillon

  • Far Hills, NJ
  • +1 908-256-2279
  • bob@the-dillons.com
Bob Dillon

Education

Education
2003 - 2004

M.B.A

COLUMBIA UNIVERSITY - COLUMBIA BUSINESS SCHOOL
1987 - 1991

BSCoE

NEW JERSEY INSTITUTE OF TECHNOLOGY

Work History

Work History
2012 - Present

Founder

ARTIMYS LANGUAGE TECHNOLOGIES

Artimys Language Technologies is a web-based service that enables online communities and messaging services to protect child subscribers from harm by analyzing electronic messages to detect sexual predators, bullying, signs of suicidal intent, sexual activity and drug use.

I'm a Founder in the most Entrepreneurial sense of the term: I do a lot of the work that goes into building the company and the product. I fill many roles from business and technology strategy to hands-on individual technical contributor. I've done all of the company's Data Science starting with a simple prototype in Octave using libSVM, followed by a complete prototype implemented on the google App Engine in Java using the Google Prediction API for Machine Learning. The production system implemented the text processing using Solr/Lucene and the Machine Learning using SVMlight and Google Prediction API.

Raw training data was scraped and collected from curated sites (Perverted-Justice), Usenet rooms as well as from researchers. Raw data was processed using custom bash and Python scripts to produce final Corpora for training. Certain data sets were classified using Amazon Mechanical Turk.

I wrote the prototype web app (www.milkmoose.com) in Java on the Google App Engine (GAE).


I specified the production system and hired a developer to implement the system. The analyzer is implemented on Amazon Web Services (AWS) and the analyzer API front-end is implemented on GAE. The Facebook app is implemented in AWS Beanstalk.

Created and ran a Kickstarter campaign to raise money for a consumer app that uses the analyzer to protect children's Android phones.

Jan 2004 - Present

Vice President, Mobile & Semiconductors

IBIQUITY DIGITAL

Developed new business and product: FM+ app. FM+ is an Android app that combines the smartphone's native FM radio receiver with IP services to deliver a modern radio listening experience. Develop product concept, roadmap, business strategy and partnership strategy. Lead business and development team to bring FM+ to market in Mexico and elsewhere around the world.

Lead development of HD Radio Guide app. App serves as a collateral promotion piece for HD Radio by informing consumers about unique content available only via HD Radio, radio receivers available and upcoming events related to radio.
Manage Hybrid Radio products. Combine HD Radio and analog radio with IP services to create a modern radio listening experience in the car. Responsible for team that deveops product roadmap and partnerships.

Responsible for licensing semiconductor partners and new semiconductor partnerships. Generate required levels of revenue and cash while signing up strategically important partners to advance business objectives.

Responsible for placement of HD Radio in mobile phones. Develop the value proposition and selling tactics. Manage team responsible for working with Mobile Carriers and handset OEMs.

Jun 2001 - Feb 2002

V.P. Powerline Networking

AWARE, INC

Developed business plan for entering the powerline broadband access equipment and technology market.  

1998 - 2001

CEO & co-Founder

ENIKIA

Enikia was a pioneering technology company that made significant contributions to the Home Networking Market. Leading developer of powerline home networking technology and semiconductors. As founding board member of the HomePlug Powerline Alliance I played a leadership role in the development, standardization and adoption of powerline home networking technology.

Led 50-person company through $15 million of fundraising, the development of constituent technology to the HomePlug standard and numerous partnerships and industry alliances.

1996 - 1998

Product Marketing Manager

DIALOGIC
1991 - 1996

Hardware and Software Engineer

ARIEL CORPORATION

Designed hardware and software for high performance multi-processing signal processing computers used in industrial and military applications.

Honors

Winner - Grand Prize - 1994 EE Times American by Design Contest - Prize: $10,000
Winner - Grand Prize - 1995 EE Times American by Design Contest - Prize: 1957 Ford Thundebird

Patents

DIGITAL RADIO BROADCAST RECEIVER, BROADCASTING METHODS AND METHODS FOR TAGGING CONTENT OF INTEREST

US Patent: US 8660479 B2                                                                                                   Issued , 08 / 2010

A digital radio broadcast system includes a processing system that receives first audio content, first program data identifying a first item for the first audio content, second audio content, and second program data identifying a second item for the second audio content such that a start of the first program data is received at the processing system within 0.5 seconds of a start of the first audio content. A digital radio broadcast signal including the audio content and the program data is processed for digital radio broadcast transmission via a transmitter. The processing system stops delivery of the first program data to the transmitter upon receipt of the second program data, the first program data thereby being truncated, and begins delivery of the second program data to the transmitter. A digital radio broadcast receiver can tag content of interest based on a user command registered at the receiver.

DIGITAL RADIO BROADCAST RECEIVER, BROADCASTING METHODS AND METHODS FOR TAGGING CONTENT OF INTEREST

US Patent: US 8,351,843 B2                                                                                                   Issued , 01 / 2013

A method for specifying content of interest using a digital radio broadcast receiver is described. A digital radio broadcast signal includes first audio content and first program data, wherein the first program data includes information identifying a first item, and includes second audio content and second program data, wherein the second program data includes information identifying a second item. A user command entered at a user interface during reception of audio content is registered, indicating a user's interest in either the first or second audio content. It is determined whether there is an ambiguity in the content of interest. If there is an ambiguity, a first data structure is stored for the first audio content, and a second data structure is stored for the second audio content. The first data structure includes the information identifying the first item, and the second data structure includes the information identifying the second item.

DIGITAL RADIO BROADCAST RECEIVER, BROADCASTING METHODS AND METHODS FOR TAGGING CONTENT OF INTEREST

USPatent:US8676114B2                                                                                                       Issued , 03 / 2014

A method for specifying content of interest using a digital radio broadcast receiver is described. A digital radio broadcast signal includes first audio content and first program data, wherein the first program data includes information identifying a first item, and includes second audio content and second program data, wherein the second program data includes information identifying a second item.

Publications

NEURAL NET NONLINEAR PREDICTION FOR SPEECH DATA                                                                                         05 / 1991

IEE Electronics Letters 

A new, nonlinear, neural network based, predictor has been devised for the encoding of speech data. It may be used in the design of a differential pulse code modulation (DPCM) coder for speech. A hybrid neural network architecture has been employed which combines the perceptron and backpropagation paradigms, thus called the PB-hybrid (PBH). Only two neurons are needed in the backpropagation section, keeping the required overhead modest. This predictor is designed by supervised training, based on a typical sequence of digitised values of samples in a speech frame. Simulation experiments have been carried out using 15 ms frames of 16 kHz speech data. The results obtained for the prediction gain show a 3dB advantage of the PBH network over the linear predictor.