- Dialog technology and NLP consultant for HealthText, a text-messaging-based platform for providing timely and accurate health information to underserved (e.g., Medicaid) populations. Supported by NSF SBIR Phase I grant.- Wrote Wizard-of-Oz (WoZ) scripts to support human-in-the-loop text messaging platform for providing health information.- Consulted on simplest technology and system architecture for building automated chatbot with minimal complexity and cost while meeting necessary requirements.- Based on WoZ scripts, developed Google Cloud-based automated chatbot to provide health information. Assessed effectiveness of platform MVP in providing accurate and relevant health information.- Wrote relevant sections of SBIR Phase I report, documenting methodology and conclusions.

At Brain, our vision is to change the way you get the stuff that you need. Our iOS app "Natural" lets you do online shopping, food delivery, groceries, and more, just by saying or typing what you need.The Brain NLP group builds the services that find out what you want, so we can get it to you! We are extremely user-focused, and use a range of methods and technologies from simple string matching to advanced deep learning methods and the most complex large language models such as GPT-3.- Pivotal in defining and managing company and team objectives and key results (OKRs).- Led company financial and expense management.- Defined product features and direction as related to NLP technology and user interface, including tight interaction with product and design teams.- Leading team building core capabilities in natural language processing (NLP) such as domain/intent classification, slot filling, query completion, and more using the latest in both rule-based and Deep Learning methods.- Designing and building data pipeline for managing data toward a process of continuous improvement of NLP services such as intent detection and slot filling.

CloudMinds is the first cloud robotics company. We are building the platform that will drive and manage our growing ecosystem of smart robots.As Director of NLP Applications, I am leading a team building core capabilities in NLP such as domain/intent classification, slot filling, question answering, information extraction and more using the latest in Deep Learning methods and tools such as TensorFlow. I am also building ASR capabilities using a mix of 3rd party service and in-house capabilities.

Independent consultant, working with clients in areas relating to:- Speech recognition- Natural language- Speaker verification- Intelligent Assistants

Responsible for speech recognition (SR), natural language (NL), advanced search, and IVR dialog management for a Intelligent Assistant platform to improve Patient and Provider experiences.NL: Developed Question Answering system and NL infrastructure for understanding Patient intent from open queries, and producing content-sensitive answers, including support for rapid NL content generation. Tools: Python, Stanford Core NLP, NLTK, WordNet and other 3rd party taxonomies.SR: Incorporated multiple SR engines into platform. Developed algorithms for best utilizing and evaluating multiple cloud and on-device SR engines. Tools: Cloud SR engines, Pocketsphinx, iOS/ObjectiveC/XCode, Python, MySQL, Amazon S3.Search: Incorporated advanced Search capabilities into platform, allowing efficient management of dynamic customized content, and content access via natural speech/text interface. Tools: elasticsearch, Lucene, Grails/Groovy, Java/Eclipse.iOS: Developed iOS client functionality for SR, NL, and search-related components. Backup for primary iOS developer. Tools: iOS/ObjectiveC/XCode.IVR: Developed framework for IVR dialog management. Integrated IVR dialog framework with existing content tools, and cloud telephony. Tools: Python, Amazon S3, cloud-based telephony platform.Product Management: Managed roadmap for various threads of product development. Worked to translate dynamic business requirements into product features, and with development team to feed features into Agile process.22otters (www.22otters.com) is a virtual health assistant platform, incorporating smart phone applications as well as SMS and voice call notifications, that coaches patients through interactions with medical providers, especially related to complex procedures. The HerStory app (http://www.22otters.com/herstoryapp) is creating a virtual community for women in various stages of breast cancer. Users can find and listen to stories in the application, and also create their own stories.

• Subject Matter Expert and principal contributor for all Speech Science-related tasks on all client engagements, responsible for all aspects of speech grammar development and performance optimization.• Perform grammar tasks on applications using Nuance OSR 3 and Recognizer 9, including use of Nuance tools for grammar testing and validation. Developed in-house tools and infrastructure for more automatically generating grammar test cases from User Interface Specifications.• Perform grammar optimization tasks on application using Nuance OSR 3 and Recognizer 9, including use of Nuance tools such as nuance-tune, NAR, and acc_test. Developed unified framework for utilizing such tools on OSR 3 as well as NR9 applications. Developed tools and infrastructure for running in-depth experiments and analysis for improving grammar performance.• Instrumented Solutions Performance methodology with marquee Clients. Worked with Client to ascertain Key Performance Indicators (KPIs) for the contact center application based on business goals. Developed custom infrastructure and tools for reporting on application KPIs, including relevant trending. Starting from KPIs and moving to task-level and grammar-level metrics, performed analysis and made suggestions for application changes/improvements, and continued to measure and report on KPIs through the various versions of the application. Continuing to support Client initiatives for application improvement by providing detailed data on performance. Suggestions for improvements resulted in significant increases in Application performance and business benefit.• Am engaged in pre-sales activities with current and prospective clients as proposed engagements related to Speech Science, Solutions Performance, or other aspects of Speech/Natural Language core technologies.• Am managing the laboratory infrastructure (Linux/Windows Server machines) needed for grammar development, grammar performance optimization, and Solutions Performance.

• Instituted a Solutions Performance process with 2 major Financial Services clients, involving deep discovery of Business Objectives for a client-facing IVR application, development of Key Performance Indicators (KPIs) in order to track achievement of Business Objectives, implementation of a technical process for measuring, tracking, and optimizing KPIs. Implementation of technical processes required custom log file manipulation, processing, and merging.• Developed novel uses for Natural Language (NL) technology in non-call-center domains for Topic Spotting, leading the effort from presales into development. This application of NL could potentially open up new market possibilities.• Was major contributor to requirements definition, design, and development of a completely new computing infrastructure and process for optimizing the performance of IVR applications, incorporating requirements for cost, performance, usability (> 100 users in various technical disciplines), and compliance/security.

• Managed a team of up to 19 Managers, Project Managers, User Interface Designers, Application Developers, and Speech Scientists, with minimal team attrition.• Maintained extremely high Client satisfaction and deep relationships with Clients. One marquee Financial Services Client noted that their engagement with our team was “one of the best” they have seen, which a top Healthcare Enterprise characterized our team’s relationship with them as “like sitting with us at the Round Table.”• Architected a complex “gainshare” engagement where significant bonuses were set (and paid) for exceeding KPI targets. This engagement was considered a significant “win-win” from both the Nuance and Client standpoint.• Acted as Professional Services Subject Matter Expert (SME) for Natural Language, actively involved with defining Product strategy along with the Marketing organization, being the PS representative through the internal Product approval cycle, managing an internal beta program for the product, disseminating Product information and best practices internally and also externally in Pre-sale engagements.

• Managed a geographically distributed team of up to 8 Speech Scientists.• Supervised development of Best Practices for Speech Science in the PS organization, including application Optimization at the grammar, transaction, and application levels.• Responsible for computing infrastructure needed to support Speech Science activities.• Acted as Professional Services Subject Matter Expert (SME) for Voice Biometrics, disseminating Product information and best practices internally, and also externally in Pre-sale engagements. Was almost exclusively the resource engaged for difficult performance issues relating to Voice Biometrics.• Was consistently requested from Marketing to speak at the annual Nuance user/partner conference on best practices for voice biometrics.• Developed tools and best practices for optimization of Voice Biometrics applications, which are still used today.

• Managed a team of up to 5 individuals performing various activities for Brokerage applications.• Supervised development of best practices and reusable assets for Brokerage applications, which were used in several deployments, significantly improving efficiency.• Acted as PS conduit with Product group developing external tools allowing clients/partners to perform Optimization on IVR applications.

• Performed Project Management, Design, and Optimization activities for over 10 IVR applications in various domains.• Specialized in advanced Speech Science tasks as needed for difficult domains (e.g., large grammars).

Voxware was developing state-of-the-art speech coding and voice transformation algorithms for games and other applications. I developed novel algorithms for transforming speech into (typically well-known) voices.

I was a Graduate Resident Tutor (GRT) for 30+ undergraduate students in the MacGregor House dormitory. I lived in the living group with the undergradate students of all years.Responsibilities included:- Ensuring safety of undergraduate living environment- Coordinating house-wide events with other GRTs and Housemasters (live-in Faculty)- Assisting with and facilitating living group events.- Yearly training on relevant issues relating to undergraduate life- Training on First Aid, CPR, co-counseling, mediation

Trained and licensed in the State of Massachusetts to act as Mediator for disputes.Engaged by City of Cambridge, MA on a volunteer basis to mediate disputes as alternative to small-claims court.

Performed research in the following areas:- Signal processing algorithms for automatic speech recognition- Advanced signal processing for speaker identificationPublications listed separately on profile.

I performed algorithm development, data collection, and system administration duties in the Artificial Intelligence Laboratory Speech Group. As part of this role, I supervised the collection of the NTIMIT (Network TIMIT) speech database (cited separately).Publications listed separately on profile.