JOSEPH W. MARIN

Directed adoption and use of technology to improve efficiency, accountability, and scalability of operations for a small but rapidly growing restaurant group. BFW Group has expanded to 15 locations across several brands, with more locations currently under construction. Expansion plans include over 100 new locations within the next two years.

• Evaluated numerous restaurant point-of-sale software providers; selected and supervised installation and configuration
• Standardized technology across restaurants to simplify maintenance and monitoring, including physical infrastructure, system hardware, and software
• Introduced new technologies to increase sales and efficiency within restaurants, including digital menus, table-side ordering, mobile payments, self-service kiosks, and integration with delivery partners
• Introduced analytics for monitoring restaurant and employee performance
• Automated inventory and purchasing management for resupplying restaurants and coordinating central distribution
• Managed system security, including physically securing restaurant hardware, protecting passwords, and complying with data protection laws (LGPD, GDPR)
• Reviewed architectural plans for new restaurant construction and specified network access points
• Migrated company from small, local email provider to Google Workspace, to provide document storage and protection, simplify internal communication, and increase collaboration
• Assisted with development and deployment of company website
• Planned strategy for managing social network account access

Contributed to design and development of features for a conversational, voice-based AI system for handling drive-through and phone orders for restaurants. System uses microservices for various stages of order processing including speech recognition, text-to-speech, NLP, intent classification, response generation, and order fulfillment. Services and APIs are written in Go and deployed on Kubernetes clusters using Google Kubernetes Engine. SQL backend uses CockroachDB, and frontend tools are implemented in Angular.

• Integrated restaurant POS systems with ConverseNow's proprietary voice-based menu format
• Developed APIs in Go (including Goroutines and Go RPC) for menu retrieval, menu changes, order pricing, and order placement
• Developed algorithms in Go for automatically translating traditional restaurant POS menus to proprietary voice-oriented format
• Contributed to development of Angular frontend for managing menu changes
• Contributed to design and development of database schema
• Managed software versions and deployments using Kubernetes clusters on Google Cloud Platform

Employee #2 and principal engineer in a technology startup building Qu (formerly Gusto), an industry-changing cloud-based restaurant management platform that includes a .NET Core API, point-of-sale software, kiosks, mobile and web ordering, catering, web-based configuration tools, and web-based enterprise reporting. Qu is installed in thousands of restaurants across the U.S., and is on a steady growth trajectory. Qu has processed hundreds of millions of transactions, totaling billions of restaurant revenue, easily handling the high transaction volume of morning, lunch, and dinner rushes.

Designed and developed a .NET Core API supporting cross-platform restaurant applications.

• Flexible, configurable data model for enterprise configuration of restaurant data (menus, employees, discounts, etc) with an inheritance model for common configuration elements across locations, implemented on multi-tenant SQL Azure database, MongoDB, and Amazon S3 storage
• Azure web services (REST) for secure data access with enterprise permission model
• Complex and flexible check-pricing algorithms to support virtually any federal, state, and local laws regarding taxes, discounts, service charges, loyalty rewards, and tax-rounding rules
• Credit card payments via multiple processors (Vantiv/Worldpay, Elavon, OpenEdge, Heartland), supporting magnetic stripe readers, EMV (chip) devices, and NFC readers (Apple Pay and Android Pay), including certification of rigorous data security requirements mandated by Payment Application Data Security Standard (PA-DSS)
• Integrated third-party loyalty, reward, and order-ahead applications, including UberEats, DoorDash, LevelUp, Clutch, Punchh

Designed and developed a Windows UWP point-of-sale application for managing restaurant orders.

• Offline data synchronization for seamless operation during internet outages
• Peer-to-peer data sharing among onsite terminals using TCP communication over Windows sockets
• Restaurant hardware integrations (payment devices, barcode scanners, Mettler-Toledo and Detecto scales, receipt printers, customer facing displays, touch-screen kitchen displays)
• Internationalization, supporting configurable currency and date formats, plus multiple simultaneous languages to allow the UI to adapt dynamically to the logged-in user’s preferred language, while displaying a location-based default language on customer-facing displays, and a third configurable language for displays and printers at kitchen and food-preparation stations

Designed and developed cross-platform restaurant applications for kiosks, mobile devices, and web.

• Restaurant ordering self-service kiosk application built with Ionic Framework, including style templates for customized branding and advertising
• Mobile-friendly web ordering application built with AngularJS
• Configuration, Management, and Reporting progressive web application (PWA) built with Vue.js
• Notifications on in-restaurant terminals for order processing, and order status updates

Contributed to development of Qu Voicebot, a Python-based voice ordering prototype.

• Integrated Microsoft Azure Speech Services
• Integrated Google Cloud Speech-to-Text
• Implemented microphone-based audio capture with audio sample format transformation
• Reviewed Rasa NLP/dialogs integration

Served in several roles with growing responsibility, including natural language research and software development. Provided technical direction and determined system architecture for a variety of APIs, applications, and data models for large-scale document and media management (storage, categorization, and search).

Designed and implemented Android applications for collection, organization, storage, and sharing of intelligence information.

• Data stored in local SQLite database, using SQLCipher 2.0.8 with AES-256 encryption
• Custom camera activity with voice and text annotation of photos, and video recording
• Lucene-based text analysis and translation of imported electronic files
• Biometric data collection (fingerprints, iris, face recognition, voice data) on custom military hardware
• Hierarchical organization of data, custom tagging, and data visualization tools
• Data sharing using REST services and bluetooth device-to-device transfer

Designed and implemented mobile web applications for document search and retrieval.

• HTML5, CSS, jQuery Mobile, and Sencha Touch (ExtJS)
• REST web services

Designed and implemented REST web services for document search, retrieval, and upload.

• Windows Communication Foundation (WCF)
• XML and JSON serialization

Designed and implemented a Windows Presentation Foundation (WPF) desktop application using Microsoft Composite Application Library (Prism).

• WPF host shell for shared functionality such as authentication
• Loosely-coupled WPF modules for document scanning, electronic document import, workflow design, and system administration. Modules employ the MVVM pattern, data binding, templates, commands and events, loose XAML for styles, and resource files for localization. Modules communicate via events and services. Application is extensible, allowing drop-in of new functional modules that are discovered and loaded automatically at startup

Designed and implemented a multi-tier ASP.NET application for complex search and edit capabilities.

• Forms authentication, out-of-process session management, WCF web services, and DHTML/AJAX (JavaScript & jQuery)
• Steered performance tuning strategy, including database performance, application-tier data caching, paging of large data sets, and asynchronous data retrieval

Senior researcher studying and building experimental language processing tools.

• Developed machine learning tools to extract and store intelligence data from a variety of data sources
• Guided transition of experimental language processing tools to production quality code
• Authored proposals for funding of research and development of speech processing strategies

As a senior developer in the Microsoft Speech Technologies group, drove all phases of development on a variety of high-profile, pioneering .NET speech technologies that served as a foundation for later technology including Microsoft's Cortana digital assistant.

• Designed and implemented Visual Studio authoring tools for Windows and web-based speech applications released with the Microsoft Speech Application SDK 1.0
• Designed and implemented a multi-threaded management console for Microsoft Speech Server, using WMI and SOAP
• Designed and implemented the Microsoft Prompt Engine, a voice prompt engine for telephony applications
• Hosted live web cast demonstrating the Microsoft Speech Application SDK to over 300 people
• Produced code samples and best practices documentation
• Served as cross-team coordinator for technology integration
• Technical mentor to junior developers

Software engineer in successful startup company pioneering speech recognition and speech synthesis technology, ultimately acquired by Microsoft Corporation for nearly $100 million.

• Developed core technology for speech synthesis software, including algorithms for automatically segmenting recorded speech into triphone source data for synthesis, and algorithms for selecting and concatenating source segments during synthesis
• Modified algorithms for prosody generation, resulting in higher user ratings of synthesized speech intelligibility and naturalness
• Hired voice talent and collected studio-quality speech data

As a postdoctoral fellow, conducted research aimed at understanding and modeling speech processing in the human brain.

• Developed theoretical and computational models of language processing capabilities of both normal and brain-injured people
• Designed and developed applications for experimental study and clinical evaluation of language processing abilities of traumatic brain injury patients. Techniques included precisely timed visual and auditory presentation of stimuli, vocal response recording, collection of reaction time data, and speech pattern analysis
• Incorporated speech synthesis technology in experimental treatment methods
• Hired and supervised research assistants
• Chaired weekly rounds