Project Manager — Android based Odoo POS

Open Source Android POS based on Odoo (formerly OpenERP) is an open source point of sale (POS) project that has been customized to take credit cards and specific loyalty and stored value cards that are in large scale use in the vending and food service industries.

CTO Services managed all the software development, credit card and loyalty card integration, and selected the hardware partners. The solution is PCI compliant and supports mag stripe, EMV chip, and PIN and NFC (Apple Pay and Google Wallet).

The system is based on an Android graphical touch screen and scanner that can be used to scan products, coupons, and loyalty/stored value cards.

Technologies used in the HipPOS system include:

• Python and Postgres running on Linux as a back end
• JavaScript running on an Android front end in a browser (the system is compatible with any mobile or desktop browser that supports JavaScript).

Challenges included:

• Finding the right developers who were familiar with Python, JavaScript and Odoo.
• Several changes in the approach taken on credit card processing, ending with the decision to use pre-approved equipment from Heartland Payment Systems that takes the system largely out of PCI scope.
• Integration with stored value loyalty systems used in the vending industry.

Partner / Part Time CTO — Rapport

Rapport is designed to help small and midsized businesses realize the benefits of sustainability: Save Money, Comply with Customer Data Requests, Win New Contracts and Bolster your Brand.

Glenn is one of the four partners of Rapport and serves as a part time CTO. The award winning Rapport.io platform is revolutionizing and democratizing sustainability tracking for small and medium sized business. Glenn manages the development of the entire web application.

Rapport uses the following Technologies

• Ruby on Rails
• MySQL database (migrating to Postgres)
• Several RoR plugins

Challenges included:

• Making the system quick and easy for first-time users to set up.
• Ensuring that the web application was functional across desktop, tablet, and mobile browsers.
• Evaluating and choosing the best dashboard and BI tools – included looking at Microsoft’s PowerBI, which was released as open source in August 2015.
• Adapting the architecture to be flexible enough to support localized CO₂ equivalency data to aid in the calculations of carbon footprints.

Co-Founder & CTO (company aquired) — VendScreen

VendScreen is a network of interactive touchscreen devices powered by a proprietary cloud-based platform. The screens retrofit on vending machines to enable payment, deliver information about products, and provide a more interactive consumer experience that is more like attended retail.

I was one of co-founders of VendScreen and served as the company’s CTO from inception in 2011 through January of 2014. We entered a vending market crowded with cashless payment options – including some heavyweights with known brands and a four- or five-year head start!

Our goal was to differentiate VendScreen from standard cashless devices and create a disruptive message in the marketplace. We told operators, “If you’re considering cashless payment, don’t make a move until you check out VendScreen.” VendScreen offered all the benefits of cashless payment with the added WOW factor of a digital touchscreen.

VendScreen was certainly one of the most interesting, comprehensive, and challenging IoT ecosystem platforms that I ever built.

The main component was a custom Android credit card/terminal device that was built to fit into the standard cutout on modern vending machines usually reserved for the payment system (bill acceptor or card reader). The first rev of the unit contained a single-board computer and a custom IO board that had power management and interfaces to various vending machine interfaces (MDB and DEX). The device also had a built-in cellular module for cloud connectivity.

The VendScreen devices “talked” to a cloud infrastructure hosted on Amazon that included a web application for configuring the devices and managing key content like nutritional data, product images, and on-screen advertising.

There were also a few mobile apps aimed at vending operators to support installing the devices and tracking service calls at machines. These apps also communicated with the cloud infrastructure.

The entire system had integration points into a credit card processor for card acceptance. In our case the devices talked directly to the processor allowing for less scope for PCI and faster transactions. Other integration points included an accounting system for billing and settlement and the vending industry’s NAMA VDI, which enabled the system to transmit sales data back to vending operators’ vending management software.

While there were many challenges and triumphs, the biggest achievement was getting the entire system from a few prototypes (12) to a production system ready to deploy (first 500 units) in about six months. Here are a few of the challenges we overcame along the way:

• Holding back on deploying a single unit into the field until we had full capabilities to remotely debug each unit using SSH and push full application-level and OS-level updates (no small feat :-). We used a VPN with fixed IP addresses in AWS integrated with our carrier networks.
• Numerous issues hardening the Android kernel and getting it ready for distribution as an embedded device (vs. consumer). Of course, that included making it hack proof.
• Working through several issues regarding spotty cellular connectivity and issues with modem firmware, the Android RIL (Android’s term for modem driver), and latency.
• Getting the system to communicate with several thousand variations of vending machines and board firmware releases that are supposed to follow common standards but never do exactly!
• Incorporating one of the smallest hidden NFC antennas into the design to support Apple Pay and Google Wallet.
• Struggling through issues with MySQL and eventually migrating to a Postgres database that was more suited to high transaction volumes.
• Configuring true “auto-scaling” in AWS to automatically spawn more servers to handle high loads during certain times of the day.

Advisor — Acoustic Stream

Acoustic Stream is a smart device that musicians can install on their guitars to record music wirelessly, tune with precision and protect their instruments by monitoring critical conditions like temperature and humidity.

This product uses BLE to monitor environmental conditions. Acoustic Stream led a successful Kickstarter campaign and also has OEM deals signed with major companies in the music business.

I provide high-level startup and technology consulting to Acoustic Stream through WPI’s Tech Advisor Network program. I was one of two co-leaders that guided the team through the program and continue to provide advisory services to the founder, CEO, CTO and development team as needed.

Partner / Advisor — Cooler Connect
Cooler Connect is a cellular-based tracking and monitoring device for ice cream coolers like the ones you’d find in convenience stores all around the world.

The device uses legacy 900 Mhz sensors for remote temperature and door sensing in the coolers. Manufacturers and distributors use the Cooler Connect device to keep track of how many times the cooler door is opened and to monitor the temperature to make sure it stays within an acceptable range. If the temperature gets too far out of range it can cause a very expensive “meltdown.”

The Cooler Connect device also sends alerts if power is lost. It has simple GPS capability using Telit’s Cell Locate. I chose the Telit 910 family chipset because of the number of different configurations this chipset comes in with different features. The system is currently in testing with one of the largest ice cream companies in the world.

I have been an owner and partner in this project since the beginning and oversaw the hardware design and cellular connectivity. I selected the chipset, set up and negotiated the carrier relationship, reviewed the contract and designed the network.

Although the design of this project was fairly straightforward, there were a number of unique design aspects that made the project challenging and exciting:

• We decided to use an internal cellular antenna with a switching jack to facilitate the use of an external cellular antenna if needed. This simplifies the installation for over 90% of coolers where an external antenna is not needed.
• The initial design used a dedicated GPS chip that was not reliable when the unit is deployed indoors. For subsequent designs we used the carriers’ cellular location technology and Telit’s Cell Locate system to get a less exact (within a few city blocks), but more reliable position with a simplified hardware design and lower cost.
• Because the sensors are installed inside the cooler and operate in low temperatures, special consideration was given to the sensor battery and chemistry.