Michael L Mehr

For UC IoT class, with Big Data feature to generate data for Apache Kafka. I designed a Raspberry Pi-based device that would take regular samples of several sensors and report them to a website. I designed the hardware, wrote the code for the website and wrote all internal scripts, mostly in Python. A MySQL database was used

The main website was designed to display the results of "weather" samples taken from my computer room (or wherever the device is located) - temperature, pressure, humidity, and light levels are reported. Samples are taken automatically every five minutes. Can also take a sample at the press of a button, change the state of an LED, and take a picture of the LED and text display of the device. The picture and last day's worth of data samples are displayed beneath the command and status buttons. In addition, RESTful Web APIs are provided for all functions. Added update automation.

API for X42WS weather station access
(UPDATE: Site taken down 2/2018)
Web Server: http://d48f7d79.ngrok.io
REST API Server: http://7c422863.ngrok.io/weather/api
Measurement: /sensors
Latest data: /sensors/latest
Last 15 minutes of data: /sensors/latest/15
Relay ON: /led/0/1
Relay OFF: /led/0/0
Take picture: /camera/0/3

This project involved setting up a basic system to simulate a Smart Light Bulb using FreeRTOS on an NXP LPC1769 (ARM M3) development board. The code is written in C.

I chose to implement all the requested features and add a few of my own. The code may be of particular interest, in that I had to port an entire library of code from C++ to C to drive the Adafruit SSD1306 graphics display.

The project link includes all documentation, including demo videos of the operation of the device and a detailed PowerPoint slide presentation of the project as presented to the class.

On June 6, 2017, I completed the one-year Embedded Systems Certificate program from UC Santa Cruz, Silicon Valley Extension, graduating with Honors. I have completed the following courses (grade earned in parens):

1 - Intro to Real Time Embedded Systems Programming (5381: A+)
2 - Developing Applications for Android Mobile Devices (29156: A)
3 - Internet of Things: System Design and Prototyping (30402: A)
4 - Internet of Things: Sensors, Platforms, Communications, and Applications (30402*)
5 - Linux Device Drivers (2470: A)
6 - Internet of Things: Big Data Processing and Analytics (30454: A)

Some of the final projects are posted on my GitHub page, including presentations.

* (This course was part of a restructuring of the UC certificate curriculum. While I did not receive a grade, I did do all the work and adapted the final project of a Raspberry Pi Weather Station for use in the Big Data course I was taking at the same time.)

Various projects submitted to GitHub from my Swift learning phase. StampCollection shows CoreData use and embedded JavaScript website data scraping. TimeZoneExplorer used a Parse back-end to implement a small user/admin system. GoldenPythag shows how the London client front-end might have looked as an iOS app. ShowTableBug is a small project (should have been a gist) that shows a bug discovered in the Apple Swift 2 compiler (since fixed).

A small excerpt from the project I worked on with a client in the financial services industry. Motivations for using this kind of mathematical tool can be found with an internet search of its title. Still, the client had specific needs that weren't being met by standard software, and I was able to provide a more generalized version that met his needs.

The code is in C++, as was the project. The client needed a large list of tools and capabilities in the code, all of which I implemented and debugged to their satisfaction. We evolved this family of tools from a Windows research app, to a client-server web app, to an automated data-producing app, for various business needs.

Developed a market research codebase for a client in London. This was first turned into a Windows MFC app that implemented the client's theories and algorithms for his own research. Implemented many (around 50-60) algorithms, many with several variants, using time series, planetary motions, and various math tools (see Square of 9 excerpt). The app was also ported to Linux as a feasibility study, the string and date classes needing some adaptation for astronomical time spans. Used a third-party library to incorporate NASA JPL planetary data.

Phase 2: Some code was adapted for use on a private WAMP server. Initially used Dojo JavaScript framework for the UI, but went with pure HTML to keep project simple. I used Uniserver to provide the basic server WAMP framework and wrote custom PHP and MySQL code to link the user front-end and database, which would run the proprietary C++ app on command. I took into account OWASP app security standards and added a small Forth front-end to obfuscate the site a bit from hackers. However, ultimately the client scrapped plans for the site before deployment.

One of the lead developers of the Summit machine-control software platform in use on the company's UV line of film thickness test products. Worked on the system for entire life cycle from object-oriented design and analysis through coding and debugging to maintenance. Kept current with industry standard development techniques and led co-workers to adopt best practices and standards in C++ with MS Visual Studio. Mentored the development team for many years, and assisted in the implementation of CMM Levels 2 and 3+, to a successful audit. Also performed sys-admin duties to automate software builds and maintain code base. Worked with other product teams (quality assurance, marketing and sales, manufacturing) to ensure customer satisfaction.

Primarily responsible for development and refinement of embedded application software to control wafer alignment subsystems of various client measurement systems. Used industry standard Cognex Patmax computer vision tools to develop a fully automated system that could be factory calibrated and later trained by its users to recognize sites on the user's various wafer patterns during semiconductor manufacture. This would allow the system to align the geometry of stage motion to precisely locate microscopic sites on wafer with reliable production accuracy and high speed. System was in production use for over a decade, and well appreciated by its users on customer satisfaction surveys.